Workshop: Nonaffective Psychosis in Midlife and Beyond - Day 1
LAURA ROWLAND: Welcome. On behalf of the NIMH organizing committee, Drs. Andrea Wijtenburg, Craig Fisher, and myself, Laura Rowland, we thank you for joining the NIMH Workshop: Nonaffective Psychosis Midlife and Beyond. So I first want to give a special thanks to Jonelle Duke and the team at The Bizzell Group for the logistical production in this workshop.
And now I will go through a few housekeeping notes before we start the workshop. So first, attendees are entered into the workshop in listen only mode with cameras disabled. If you want to learn more about today’s speakers, I invite you to go to our registration website where you can find their biographies.
We encourage people to submit questions. Please do this through the Q&A box, and please address the question to the specific speaker that you have a question for. If you have any technical difficulties hearing or viewing the workshop, please note this in the Q&A box and our technicians will work to fix the problem. You can also send an email to Events@OneSourceEvents.Com. Please note that the workshop will be recorded and will be posted to the NIMH website for later viewing. And now it’s my pleasure to introduce the chairs, Dr. Ellen Lee from the University of California San Diego, and Dr. Sophia Frangou from University of British Columbia and Icon School of Medicine at Mount Sinai. Thank you so much.
ELLEN LEE: I am just going to go over the workshop overview for the next two days. Today we’ll be talking about the phenomenology, course, and outcomes of mid to late life psychosis. We’ll talk about cognition including social cognition, and then we’ll have a session on regulatory and metabolic processes.
All of these sessions will include short talks as well as a panel discussion and time for Q&A. Tomorrow we’ll have day two where we’ll touch on the topics of accelerated aging, treatment targets and intervention development, as well as followed up by a synthesis of the two days of information that we’ve received, and the discussion of opportunities and next steps for research in this field.
And the goals of this workshop are to share the latest findings focused on first episode nonaffective psychosis during mid to late life, discuss the phenomenology and the course of nonaffective psychosis across the lifespan, discuss the trajectories of clinical symptoms and cognitive functioning, as well as underlying mechanisms, including brain circuitry, regulatory and metabolic processes. We’ll also touch on accelerative biological aging, premature mortality for this group, as well as treatment targets and the development of novel interventions. And we want to discuss the challenges and opportunities for research in mid to late life populations.
Some of the questions we’re hoping to discuss further during these sessions include what is unique about individuals with later onset psychosis, are the symptoms and functioning distinct from individuals with earlier onset psychotic illness, and what are the trajectories of clinical and cognitive functioning across the lifespan? Is there an increased risk for dementia or other disorders or risk factors across the lifespan?
And what are the key regulatory and metabolic processes that underlie psychosis across the lifespan? And here we’ll be looking at sleep, as well as mitochondrial functioning, as well as brain biogenics, and what is the key evidence for accelerated biological aging in the periphery and within the brain itself? Could these underlying mechanisms be potential treatment targets? And are the brain and the periphery linked in terms of these biomarkers?
And last, which treatment targets and interventions show the most promise for mid to late life psychosis? What is the state of the evidence, and can these approaches be personalized for this very heterogeneous group?
And some of the other discussions will revolve around challenges of studying this population, what are the key barriers to translating what is known about nonaffective psychosis to older adults, how do we account for the complexity of aging processes, as well as the environmental and treatment influences in this population, what are the key knowledge gaps for understanding and treating nonaffective psychosis in mid to late life, and what future research directions and approaches would best address those gaps.
SOPHIA FRANGOU: Thank you. I would like to join you and Laura in welcoming everyone to this workshop, which we hope will be both enjoyable and informative. Dr. Joshua Gordon, the NIH director, could not be with us today, but has provided a video sharing his thoughts on the importance of our webinar.
JOSHUA GORDON: Welcome, to the 2022 NIMH Virtual Workshop on Nonaffective Psychosis in Midlife and Beyond. I’m Joshua Gordon, Director of the National Institute of Mental Health, and it’s my pleasure to join you here today virtually. Nonaffective psychosis affects about 20 million people worldwide and is one of the top 15 leading causes of disability. Those living with nonaffective psychoses on average have a shorter lifespan and greater morbidity than the general population from a variety of health concerns. The NIMH is committed to transforming the understanding and treatment of nonaffective psychosis, including schizophrenia spectrum disorders, through research paving the way for prevention, recovery, and cure.
For example, we’ve recently started the accelerating medicines partnership program for schizophrenia in concert with a number of private and public partners. The goal of this initiative is to identify biological markers, clinical endpoints, and other measures that predict disease trajectory and outcomes, and to generate tools that will improve the success in developing early-stage interventions for patients who are at risk for developing schizophrenia. But this can’t be the only thing. A quarter of cases with schizophrenia spectrum disorders do not fall into a clinical high-risk category early in the disorder.
The majority of people living with nonaffective psychoses are over 35 years old, including those first diagnosed later in life, and those who age with the illness. This large portion of people living with nonaffective psychoses in midlife and beyond have the highest disability adjusted life years, indicating the highest burden of disease. This highlights the need for research that focuses on this midlife and beyond population.
Detailing NIMH’s strategic plan, our goals of understanding the trajectories of mental illnesses across the lifespan, of striving for prevention and recovery and cure, and in strengthening the impact of NIMH supported research, all converge today in this workshop. NIMH is committed to better understanding nonaffective psychosis that emerges at midlife and beyond, as well as those who age with illness, and with a goal of preventing and curing.
This workshop will start to address the following. This first episode nonaffective psychosis that emerges late in life. The distinct group is this late onset group distinguished by different phenomenologies, etiologies, or pathophysiologies. What are the underlying mechanisms of the trajectory and clinical symptoms that occur during the mid to late life course? Is menopause a period of vulnerability? Are there sex differences. What are the underlying neurobiological mechanisms for the increased risk of dementia seen in those with nonaffective psychosis? Is there an accelerated biological aging process that could help explain premature morbidity and mortality that we see in this group?
And can we identify treatment targets to develop novel interventions that target nonaffective psychosis later in life. These and many other questions will be answered over the next couple of days. I’m really excited to learn of the outcomes that will help guide future research at NIMH and throughout the research community. Thank you for joining us.
SOPHIA FRANGOU: So our next video provides an invaluable perspective from a lived experience point of view, as we see Dr. Roberta Payne and Dr. Andrea Wijtenburg in conversation about this important topic.
LAURA ROWLAND: First of all, I would like to thank you so much today Dr. Payne, for being willing to speak with us today about your lived experience with schizophrenia. I guess just getting right into the questions, could you give me an idea of what your history of schizophrenia is?
ROBERTA PAYNE: It was hard to settle in on it because I was prodromal for so very long. I had my first hospitalization at the age of 22, and I was finally diagnosed with my second hospitalization at 42. I was diagnosed with paranoid schizophrenia. But before that I had delusions, terrible delusions that I had to walk in alleys at night or I might be killed by the alien beings, things like that, before I was ever diagnosed. And I had all sorts of, I had to do this exactly and do that exactly and do that exactly.
And I had paranoia the whole time, lots of anxiety, and I was really relieved to finally be diagnosed because I didn’t know what the heck was going on for about 20 years. So my symptoms when I was hospitalized were terrific paranoia, it would never leave me, and it wasn’t to leave me for another 20 years. And delusions, mostly about alien beings and evil ones and so forth and so on.
And an intensity that is very hard to describe. I was exhausted from the intensity of thinking all the time. And really screwed up thinking, I couldn’t even wash my clothes because I couldn’t figure out the meters on the washing machine, I couldn’t do any of that stuff. And as I said paranoia was the worst.
LAURA ROWLAND: Roberta, Can I ask you, have your symptoms changed with aging?
ROBERTA PAYNE: Very much so. See, I was before Clozaril and they just had very old-fashioned medications then. I was on Stelazine for a while, things like that. With Clozaril everything changed, and then the second time it changed was with Abilify, and I’m on Abilify now. The only symptoms that I have now are very minimal paranoia, but it’s always there, I always have to work on it. And hallucinations, I hallucinate voices from the shower, and the air conditioner, and in the hallway outside my apartment and so forth. But they’re really not that bad. I can’t complain about my life now compared to what was before.
LAURA ROWLAND: Since you were diagnosed, have there been highs and lows, or do you feel like it has been rather stable once you found a medication that worked for you?
ROBERTA PAYNE: Well, I found the medication that worked for me only about 15 years ago. So before that, I went through a long period that I call my mud years, because, starting about the age of 42 or 45, I felt like I was encased in mud so high I couldn’t even reach the top of it with my hands; that reflected my thinking, my inability to think properly. And the paranoia of course. Lots of panic attacks and so forth.
And that coincided I think with the menopause years. It was so bad, my illness was so bad, I don’t remember the menopause at all. The only thing, what’s interesting about it is after the mud years were gone I had no more desire to have children and I had no more sex drive, I haven’t since. But as far as the menopause goes I don’t know.
LAURA ROWLAND: And then can I ask you, Roberta, what kind of coping mechanisms have worked best for you?
ROBERTA PAYNE: I have quite a few. One has been cognitive behavioral therapy. That has helped a lot. I still use it a lot. Another has been art, drawing, and writing. I started writing about 15 years ago, and that has helped tremendously, because you have to think properly to write.
And then another thing that really helped me was my psychiatrist asked me to join the board of directors as a consumer member of the Mental Health System of Denver. And so I was on it for 13 years. They showed me how to think properly, how normal, healthy people think and solve problems and react to them. And I watched them like a hawk, learning how to think properly.
Another thing that helped was having, and I really believe in this, was having mentors. Dr. Deborah Levy of Harvard and Dr. Robert Friedman of CU were my mentors. Dr. Friedman still is, Dr. Levy of course has passed on, and we miss her a lot. But I still have a mentor. Again, they aren’t psychiatrists of mine, they’re just there as people whose thinking helps me a lot. That was about it.
LAURA ROWLAND: Have these coping mechanisms you think changed over time, or has it been pretty uniform, like this is what has worked for me for decades?
ROBERTA PAYNE: This has worked for me for decades, but I also used, in the early years, especially the mud years, I used, what do you call it, when you ask someone if something is, reality testing, I would use a lot of reality testing. I would ask people do you hate me. Are you angry at me? Have I done something wrong? I’d ask it over and over and over.
Nowadays, when I get paranoid, I don’t have to ask people, I just say, give it time, it’s not like it used to be, it will settle down. Because my paranoia would not leave back when it was really strong, it would be there for 10 years if I didn’t do something about it. But nowadays it kind of settles in, and I realize, oh, no, that wasn’t you after all.
LAURA ROWLAND: That’s good to hear. Roberta, are there any unique challenges to living with schizophrenia as an older adult?
ROBERTA PAYNE: Yes. I find that, I have a limited amount of stress that I can take. This goes back to coping mechanisms. I avoid television, I avoid rock music, I avoid highways, Costcos, things like that. Nowadays I just avoid other stressors that have come up in old age, such as health. Health has, recently, I had a rotator cuff problem, then I had a problem with my eyes which is very severe, and it's being taken care of now.
But I had too many things going on at one time, and I have learned to say, okay, I have realized I am overwhelmed. And so I’ll take it one problem at a time. But I think older people have too many problems with their health to be able to deal with them and schizophrenia at the same time, and I think older people, especially those of us who are on SSDI have very limited incomes.
And it’s hard, especially with inflation now on food, it’s very hard to live within my budget, it’s very hard, and I imagine other people with schizophrenia have basic poverty problems. So, I would say lack of money and the illness are the big problems for people who are aged with schizophrenia.
LAURA ROWLAND: Did you have these challenges when you were younger, or do you think the challenges when you were younger are different than when you are older?
ROBERTA PAYNE: There were some things that were better when I was younger, but right now I’m on a fixed income and that’s all there is to it.
LAURA ROWLAND: Roberta, are there any services or therapies that you wish were available to help you?
ROBERTA PAYNE: Not really, I think I’m getting everything that I need. I don’t know that other people are getting everything they need. I’ve been very fortunate to have all the people who have helped me. I have an army of people who have helped me. I’m just so fortunate, but I know other people don’t.
LAURA ROWLAND: So that support system is very important.
ROBERTA PAYNE: Very, very important, yes.
LAURA ROWLAND: Another question I have for you, and this was just the final one, is what are some of the positive aspects of your illness?
ROBERTA PAYNE: The illness itself had no positive aspects, but there were things that came out of it that were positive. One of them was I learned how to forgive, because there’s a lot of stigma involved in schizophrenia, and I’ve learned how to forgive people who have walked away from me. And there have been a lot. People who didn’t understand, or people who just said I can’t deal with this, I don’t like her anymore, good luck, goodbye, including my family.
The other positive thing that came from it was my thinking was so compromised during my worst years, and as I was getting out of them, I would say about the age of, oh, 55, around in there, I was just getting a pretty good hold on being able to think well, and for reasons that aren’t relevant my father came to live with me for two years then, and he had Alzheimer’s.
And as I was coming into good thinking he was descending into Alzheimer’s. And the good thing that has come from that is I have this fabulous, fabulous admiration for the human brain when it’s working properly, it’s just marvelous.
LAURA ROWLAND: I agree completely. That is all the questions that I have for you. Is there anything else that you wanted to share with the colleagues who will be watching this, or do you feel like you’ve shared your story with us?
ROBERTA PAYNE: I think I’ve shared it pretty much. Let me just add that there is a reason that people with schizophrenia are tired all the time. It’s because it’s such a demanding illness. And I totally believe in the science. And thank you all for what you’re doing.
LAURA ROWLAND: Dr. Payne, thank you so much again for being wiling to sit down and chat with us today, to share your experience with schizophrenia with us. I really appreciate it, thank you for taking the time.
ROBERTA PAYNE: One last thing. Read my memoir about schizophrenia. It’s available on Amazon, it’s called Speaking to my Madness: How I Searched for Myself in Schizophrenia.
LAURA ROWLAND: Thank you so much.
LENA PALANIYAPPAN: Okay, thanks for that. Welcome everyone for session one. I am Lena Palaniyappan, we have the first session on phenomenology, course, and outcomes. I’m going to introduce the first speakers, and I’ll just remind everyone who is speaking that we have 15 minutes to talk, and any questions the audience have can go on the Q&A box. All questions will be answered in the ten minutes that follows all four talk sessions.
Let me start with Dr. Carl Cohen from SUNY Downstate Health Sciences University. We also have the second speaker, Dr. David Castle, from the Center for Addictions and Mental Health in Toronto. We have Dr. Antony Ahmed at Weill Cornell Medicine and Dr. Iris Sommer from University Medical Center Birmingham. Over to you Dr. Carl Cohen.
CARL COHEN: So, hi everyone. And thank you NIMH for doing this yeoman task of assembling everyone for this very important topic. Today I’m going to talk about course and outcome in older adults with schizophrenia, which is the focus of my work in this area. I have no disclosures, and I’ve received funding from the NIH for some of the data that I’m going to present today.
I wanted to first pay homage to the first conference the NIH held on this topic almost 40 years ago, led by the late Gene Cohen and Nancy Miller. A number of luminaries were there, I was a younger man then, and it was a privilege to hear these talks. Gilford eventually put together this material in a book, so if anyone wants to read some classics, they should try to find the book somewhere.
So let me begin with a few preliminary things that I wanted to say. First, schizophrenia as you all know develops in the second and third decades of life, but the number of patients surviving into old are markedly increasing, even with increased mortality rates.
The prevalence estimates for people 60 and over from a study in the Netherlands found that roughly 0.5 percent of the population is 60 and over and has schizophrenia, and the ratio of early onset to late onset is about two to one, as early onset people have died off late onset people have taken their place.
The other thing that’s very important is that the number of people with schizophrenia 55 and over has been increasing markedly and will reach about one fourth of the persons with schizophrenia by 2025. By 2060, not so far off really, one half of the schizophrenia population will be 55 and over. Globally the number of persons 60 and over will double over the next 35 years and reach about 10 million people.
Sadly, only about one percent of the schizophrenia literature is devoted to older adults. I also wanted to emphasize that we’re now talking about a population that’s in the community, 85 percent live in various levels of supported care or independent living of family and the community, only 10 percent in nursing homes and five percent in hospitals.
Another point I wanted to make is that the traditional view of schizophrenia in later life has been one of stability, quiescence, and stable end stage. We can see these from prominent researchers in the field. This end stage has seen sort of a flat line almost. We want to address that today.
I’m going to present six points about outcome in older adults with schizophrenia. The first point is outcome changes historically because of evolving diagnostic criteria, criteria for outcomes and social factors. Also whether studies are conducted cross-sectionally or longitudinally.
And it’s very important I think, maybe emerging out of this meeting, to begin to reach some sort of consensus on what we mean as a phenotype of schizophrenia or late onset psychosis for that matter, and appropriate outcome measures. Also, when we’re working with older adults, what is the cutoff age, and what do we mean, older adults are 50, 55, 60, whatever.
This is a slide that Heggerty put together in the late 1990s, that showed how the percent improvement in schizophrenia historically has really changed in the early part of the 20th century, the prognosis was considered very poor. Then it began to increase, especially in the 70s, the 60s and 70s. Then there was a drop-off in the 80s and 90s, in part because DSM-III came out, but also there are historical factors as well, in how society has been willing to allocate resources for people with schizophrenia, and there’s a different change in optimism based on that.
So the view for the first eight decades of the 20th century was rather negative. Kraepelin, very poor prognosis, the dementia praecox. Bleuler also thought there was a direction towards loss of functioning, and the DSM-III, in 1980, said the most common course was one of acute exacerbation with increasing impairment between episodes.
And something changed in seven years when the DSM-IIIR came out, some of it had to do with the awareness of these long-term studies that began in the 1970s in Europe and then some more recent studies in the United States in the late ‘70s and mid ‘80s. What these people found in these catamnestic studies where they reinterviewed people 20 to 37 years after initial hospitalizations, about 50 percent of people had significant improvements or recovery. The same for social recovery as well.
So this was quite earth shattering at the time. This is Luc Ciompi’s work, and it gives you an example of how there’s so much heterogeneity in schizophrenia, and you can see that there are acute courses, chronic midlife courses, mixtures of hybrids of these things.
And then he took a kind of snapshot, and he called it interestingly end state, and found that 50 percent of the people had favorable outcomes. And many of them, it didn’t depend on their mid-state situation. So to summarize the historical trends, between 1900 and 1986, right before the DSM-IIIR came out, the prognosis was rather poor. I made it about 25 percent, maybe give some symmetry to this graph. It was considered poor.
Then the cross-sectional studies came out, and the findings suggested about 50 percent of people improved or better. And then more recently studies that we did in New York City, and another study in the Netherlands by Lang et al suggested that about 25 percent really only have persistent recovery or remission.
Point two was that we need to have some common consensus about what should be included in the outcomes. The DSM V appendix on dimensions, it’s quite instructive, and it probably should include positive and negative symptoms, depression, cognitive functioning at a minimal.
Social function should be included, community or social integration, activities of daily living, quality of life are factors that we should consider. Often, we combine the two, clinical and social functioning, and talk about recovery. And finally positive aging should be looked at, successful aging based on health, social, and psychological wellbeing.
Briefly I’m going to give you some data from our sample in New York City which consisted of 249 persons age 55 and over with early onset schizophrenia spectrum disorder. We were able to follow up on 104 people who were interviewed a mean of 52 months later, and four fifths of the persons fell within the 40-to-60-month follow-up range. We also had a community comparison matched by age and gender and race of about 113 people. The third point I wanted to make here is the outcome criteria are largely independent of each other.
So from that study we looked at things like remission, community integration, CSD, depression scale and dementia rating scale for cognitive functioning. You can see that most of the correlations are on a small to medium size level, very few what might be considered high correlations.
And if you multiply these to get the shared variances, square them, you get a shared variance of 0 to 22 percent. So there’s some overlapping, but not as much as one might expect, and they’re largely in many ways independent of each other. And this has implications in terms of treatment. Treating one area may not necessarily impact on another area. And I’ll get into that shortly.
The fourth point is the outcome is not quiescent or stable in later life but continues to evolve. So this is the classic picture of schizophrenia with a kind of flatline in the later life. This is not a correct view. What we found in these outcome domains, if you’re looking at remission, community integration, recovery, cognition, depression, is that anywhere between 25 and 40 percent of people showed fluctuation over 52 months, and some people got better, some got worse. But it’s more of a tumultuous time than people initially believed.
The fifth point I wanted to make today is outcome is heterogeneous, with a variety of combinations that necessitate a personalized approach to care and a more nuanced approach to research. So in this slide we looked at five outcome measures. We looked at remission, community integration, depression, cognitive function, and health, and we used self-health as a reasonable proxy of actual health. And we dichotomized the groups into persistently unfavorable or worse over time and persistently favorable or better over time.
And what we found is the possibility of 120 different combinations. Many people had two, three, or four of these favorable outcomes, but there were people who had none, just under 10 percent, and there were a few who had all of them. But many people had different combinations. And again, it suggests a different approach to each patient, the more personalized approach, and also suggested our research has to be a little bit more sophisticated in terms of looking at these various potential outcome measures.
Final point is recovering, or sometimes known as recovery, can be assessed empirically, and yields a five-tier taxonomy of varying degrees of recovery that provides guidance for treatment and research. Our definition of clinical recovery included symptom remission based on the schizophrenia working group criteria score of three or below on each of eight symptom domains of positive and negative symptoms on the PANSS test scale.
The community integration scale, which we had created, you need a score of nine or more out of 12 items. And these can include various social integration components and included things like social networks and activities in the community, satisfaction with life, supports, things of that nature.
As you can see from this slide, you can see that cross-sectionally we found about half the people achieved clinical remission, which is very consistent with earlier studies, but only 25 percent maintained their clinical remission over time.
Likewise, the community integration, 37 percent cross-sectionally, but only 26 percent longitudinally. We combine these two groups into the clinical recovery measure, 22 percent showed clinical recovery at one point in time, but only 12 percent longitudinally.
And positive mental health was only two percent in this population, but I should add that it was only 20 percent in the community comparison group, which was a relatively poor population. So our studies were fairly similar to younger population, remarkably similar, a little bit higher than the rates found in the San Diego study, and it was fairly similar to some of the catamnestic studies.
So here’s a taxonomy that we developed to look at clinical recovery, and we found five group. Tier one, which is the top group, is the stable state, people are in persistent clinical recovery. This is a group that we should certainly of course maintain some sort of current treatment, or perhaps consider reducing medications. Then there’s another group, 23 percent, that fluctuated between clinical recovery and non-recovery, and with appropriate interventions these persons might be able to attain persistent clinical recovery.
Then there was another group, only 11 percent, that either attained persistent clinical remission or persistent community integration, but not both. And then there was, the largest group, a fluctuating state of people who attained clinical remission or community integration at only one point in time. Clearly these people required more intensive work, but could possibly improve.
And finally, there was a stable state of 18 percent where these people never reached any form of clinical remission or clinical integration. It’s hard to even consider these people exactly recovering. And these people would of course require the most intensive types of treatment.
So the study demonstrated the importance of tracking people over time and the need to examine subgroups rather than looking only at mean values. So I’m going to end here with a thought that perhaps the secrets of schizophrenia will be found in later life, when it has reached its most develop and complex forms. I have some readings; I have some papers we’ve done.
I certainly recommend this book we put together, this is Paul D Meesters, and it’s published two years ago by Cambridge Press, which has a number of chapters on various topics in schizophrenia and psychosis in later life. So, thank you everyone. I’m going to now turn over the speaker’s podium to Dr. David Castle.
DAVID CASTLE: Thank you very much Carl for setting up the scene. Carl has given us a very nice overview of some very important areas and introduced some good terms like catamnestic. I haven’t heard that for a while, and of course something about Kaeppeler and Bleuler, even Ciompi. So we shouldn’t forget all of these people. And congratulations to Carl on your longitudinal work, which is so important in terms of understanding the course of schizophrenia.
What I’m going to do is talk a little bit about history, a little bit about epidemiology, a little bit about phenomenology, and then a bit about risk factors. And in the mix of all of that will be discussion around sex differences as well.
So Krapelin we’ve already heard about of course. People misconstrue Krapelin a little bit and say dementia praecox is schizophrenia. Well, of course it’s not. Kraepelin described an early onset severe male preponderant form of psychotic illness with usually a poor outcome, and he also acknowledged that there were other forms of nonaffective psychosis, including what he called the paraphrenias, which had a later onset and a better outcome, more females, and is probably akin to what used to be the paranoid subtype of schizophrenia, with better presentational affect and better overall outcome.
So Martin Roth in 1955, in a seminal paper, took the paraphrenia term, put late in front of it, and referred to a very late onset group, over the age of 60. And you can start seeing the sense here about early onset, middle onset, and then very late onset.
And one of the things Marin Roth showed was that not all people presenting in very late life, by the way 60 is not very late life anymore, as I get older, I redefine it, but over the age of 60 not everybody presenting with psychotic phenomena has dementia, which used to be thought, he showed the outcomes for people with late paraphrenia was much better than those for dementia.
And then of course the DSM III did a weird thing in 1980 and just sort of said, well, you can’t have an onset of schizophrenia over the age of 45. And although they relented and changed that in future editions, it really did pervade the thinking for many years, and still I think. So congratulations for Ellen and Sophia for putting this together, because I think these later onset forms of disorder have really been neglected.
Peter Rabins wrote a nice paper in 1984 saying can schizophrenia begin after the age of 44, well yes it can. And Dilip Jeste who is on this panel and one of the doyens of the field did a nice review showing of course you can have later onset forms of disorder.
And then my very good friend and colleague Rob Howard convened a conference which was one of the funnest conferences I’ve went to in Leeds Castle, which is this image with the swans, and he managed to get a whole wing in Leeds Castle for a whole weekend. We all debated whether late onset schizophrenia existed, and if it did what would we call it. My highlight of that meeting was Nancy Andreasen showing us wrestling halls in the library of Leeds Castle.
Anyway, Nancy said I don’t believe in late onset schizophrenia, and has a chapter in subsequent book entitled that, but everybody else sort of agreed that there is a late onset form of disorder or number of disorders in later life, and the very late onset late paraphrenia was rephrased as very late onset schizophrenia and psychosis. So that’s one bit of history.
I would also like to talk a little bit about sex differences. The long and the short of it is that this is one of the keys to understanding differences between early and late onset is understanding differences in sex. So the fact is that pretty much every study which has ever been conducted has shown that late onset schizophrenia and very late onset schizophrenia were essentially more female preponderant.
So I’m going to rather prosaically share some of my own research. The first lot of research will be from the Camberwell Register Study. Camberwell is in southeast London, this is a map of southeast London, the red dot in the middle is Maudsley hospital, and the surrounding area is Camberwell.
John Wing setup a register which tracked every first contact with psychiatric services over a 20 year period, and we were able to go back and re-diagnose all of those with schizophrenia-like disorder across any age of onset, and we were able to re-diagnose them according to various sets of criteria using Peter McGuffin’s OPCRIT checklist, and showed in fact that if you look at here the broadly defined but also the DSM-IIIR, which is a fairly stringently defined set of criteria, people in later life did meet these criteria, or with later life onsets did meet these criteria, so that’s myth number one that schizophrenia only onsets early.
And myth number two was the thought that females were sort of protected because of estrogens largely in their early life, and that pushed the distribution curve for females to the right. Here you can see I think pretty convincingly that in fact female and male distributions of onsets are very different, they’re not isomorphic.
Males have dramatic early peak, that’s neurodevelopmental paraphrenia schizophrenia, to my mind anyway, and then later onset there is a midlife peak. Not just female by the way, it happens in males too, so it’s not menopause I don’t think, and then there’s very late onset which is essentially a female disorder. And when you subject that to admixture analysis you can see that it’s very clear that there are different peaks in males than in females.
Turn to Australia, Australia is a large island in the bottom righthand corner of the world, just to the left of New Zealand. I lived there for a long time and was privileged to be engaged with the Study of High Impact Psychoses, which is a representative study of people with psychotic disorders ascertained in 2010.
And one of the things I’ll show you yet again, and you’ll get bored with these slides after a while, is to show that males and females can onset late in life. We didn’t include very late onsets just because of the methodology of the study, but you can see the midlife peak.
A different dataset altogether now, this is from the catchment area I service, which I worked in in Melbourne, St. Vincent’s catchment area, and we had a first episode service which took any age at onset. And you can see that indeed people could onset in later life, and I would point you to the need of somebody presenting for the first time in midlife with a job and a family is very different from those presenting for the first time in early life. So early psychosis services really need to be aware of this and respond to them.
Just to share some other Australian data, these are from John McGrath’s group in Queensland. Queensland is a state of Australia. And this is a sample of 10,000 people showing again male and female curves being quite different.
Phenomenology is also another key. So epidemiology and phenomenology and risk factors are the three themes I’m going to talk about today, and they’re all key to understanding the field. The fact of this slide is this is looking at very early onsets of under 25 versus very late onsets over 60, this is from the Camberwell sample, showing essentially that early onset disorders are much more likely to have negative symptoms in particular, and I know we’re going to have a talk later from Anthony Ahmed about negative symptoms I’m presuming, and they just don’t happen in the late onset or very late onset. And neither does formal thought disorder, interestingly.
If people with late onset definitely have very florid delusional sets, often widespread, often organized delusions, very common have hallucinosis in numerous different modalities, including olfactory, and one of the interesting phenomena is of a phenomenon called partition delusions. This is the thought that something or someone gets through an impermeable object and gets into your apartment for example. Rob Howard shared this slide with me, which is British constabulary actually making this partition delusion come true. But why should there be these differences in phenomena? Really as I said a key to our understanding of neurobiology and treatment.
Risk factors, we know that there are genetic risk factors for late onset schizophrenia as there are for early onset schizophrenia, but we showed in a family study in the UK with Rob Howard, also genetic loading in later onset for mood disorders. And also other factors, such as sensory impairment, Phillip Jeste did some really good work on this, showing it was mostly uncorrected sensory impairment. Also often these people had a rather paranoid personality structure, social isolation which went along with that, but mostly were occupationally very well-functioning.
This is again the Camberwell register sample comparing very early onset with very late onsets in terms of risk factors, and you can see family history more common in the early onsets, obstetric complications, developmental delay, other neurodevelopmental type markers, poor premorbid work and social adjustment. Marriage of course went the other way, that might just be because people had more chance to get married if they had a later onset.
These are more recent data from the SHIP study, and this shows that youth onset under 26, middle onset between 26 and 40 and late onset, and the very short sum of it shows later onset, more females less likely to have a family history of schizophrenia, more likely interestingly to be migrants, although that might be a reflection of the structure of migrants to Australia. And much less likely to have premorbid functioning, but less likely also to have substance use problems. And less likely to have had a premorbid unemployment.
The final bit of this is from a very recent paper, also using this study of high impact of psychosis samples, and what we’ve done here is looked at predictors of early onset in females and males, and they’re rather different. So females more premorbid social adjustment years, family history years, but in males you can add into that premorbid IQ, cannabis use, and poor premorbid work adjustment. I don’t have time to talk about treatments, just to say that there are treatments.
This is Rob Howard’s ATLAS study showing amisulpride being effective, but a point I’d like to leave us with is that part of the really important component of treatment is the engagement of people, people who often have had a paranoid disorder for some time, very difficult to engage, and we’re very interested now in acceptance commitment type therapy as a potential way of engagement.
My final slide is that schizophrenia as I’m sure we would all agree is not a disorder at all, it’s fundamentally heterogeneous in terms of phenomenology, I believe there are obviously different pathways to different disorders. Late onset itself is likely to be highly heterogenous. Some of them do progress to dementia but most don’t. We need to understand all of this, and always have the epidemiological and phenomenological context in our minds. Thank you very much for your attention.
I now introduce Anthony Ahmed from Weill Cornell, and very pleased to let you all know that it’s his birthday, he’s turning 23 today, his career has been amazing how much he has produced in such a short life so far. Anthony?
ANTHONY AHMED: Thank you. I have the pleasure of talking about negative symptoms. Negative symptoms have a long legacy in schizophrenia, long recognized that’s a core seminal feature that contribute a fair amount of disability as much as and probably more than a lot of other dimensions of symptoms of schizophrenia. Some of the work that we’ve done with Gregory Strauss and Brian Kilpatrick, we found that negative symptoms have both a categorical and dimensional aspect to them, and I’m going to talk about how this is relevant to older adults.
When you think categorically about negative symptoms, there’s the famous distinction popularized by Dr. Will Carpenter about the differences between primary versus secondary negative symptoms. Primary negative symptoms are idiopathic to the illness, they are central to the core of the psychopathology of the illness.
Negative symptoms emanate from other causes, like anxiety, depression, psychotic symptoms, medication effects, and so on. I like this figure from a paper by Chris Correll that classifies some of the many different factors that can contribute to secondary negative symptoms. Another classification that’s in the literature by Robert Buchannan, making a distinction between persistent versus transient negative symptoms. Again, a focus here on the enduring nature of negative symptoms.
Now, the issue with the persistent negative symptoms versus transient negative symptoms classification is that persistent negative symptoms will probably be a little more heterogeneous, including some patients who may have primary kinds of negative symptoms, and some patients whose negative symptoms are probably secondary but just happen to be persistent. The gold standard measure for capturing patients with a kind of deficit, a kind of presentation of negative symptoms, is this schedule for the deficit syndrome that was published by Brian Kilpatrick.
Now, in the absence of the schedule for the deficit syndrome you can capture negative symptoms with other common clinical rating scales, by computing something called a proxy for the deficit syndrome, which the proxy competition is basically looking at clinical scales, like in this case the primary study was based on the BPRS, where it was able to identify deficit cases by looking at the relative prominence of in negative symptoms blunted affect in relation to other symptoms, like anxiety, depression, guilt feelings, mostly emotional distress kinds of variables.
And in 2007 Goetz and callings extended this to the PANSS, where they were able to again establish that you are able to capture people with a kind of deficit kind of negative symptoms. Again, people with a deficit form of prominent negative symptoms relative to other symptoms, their symptoms are persistent, and their symptoms are persistent and prominent and enduring over time, as well as been primary.
Now, the issue with using the proxy classification scheme to try to identify people with this kind of negative symptom is that you almost always, if you don’t have the schedule for the deficit syndrome you almost always don’t know what the prevalence of the deficit kind of presentation is in your sample. So if you’re computing the proxy score, you have the efficiency of that score to actually help you identify deficit patients really depends on the base rate. If your base rate is low your efficiency is a little poorer, versus if you had a 50:50 split between deficit versus non-deficit patients.
However, despite the limitation, of course another limitation is that when you have cases that are tied in the proxy score, you run into the challenge of trying to figure out whether they all belong to the deficit category or they all belong to another type of negative symptom. So despite that the proxy score can still give you a very good indicator of how prominent negative symptoms are. So I’m going to show what it does in the context of older adults as a computational score.
Now, there have been several questions, lots of questions about the prevalence of negative symptoms in older adults, and Dr. Cohen has written a lot of fabulous papers about this throughout his illustrious career. The range is all over the place. The prevalence is all over the place. And ranging from about 25 percent in some samples to as high as 88 percent. In the WHO multinational follow-up study there was a prevalence of about 23 percent, with an average age of 51. And again, we get a range of prevalence estimates.
And then in the literature there have been questions about to what extent negative symptoms are dominating late life in older adults. And of course, you find a lot of disagreements in the literature about that.
The Putnam and Harvey study from 2000 suggested that there is a greater prevalence of negative symptoms in geriatric patients. A paper by Dr. Cohen found differences between older adults and younger patients in the severity of negative symptoms. And then there’s a question of well did those symptoms in older adults worsen over time or did they remain stable, and then there are also differences in the literature about that as well.
This figure from Mosolov and Yaltonskava’s paper kind of looks at like overtime sort of depicts overtime their view on the course of negative symptoms, meaning that negative symptoms from the prodrome increases and remains relatively stable, in relation to psychotic symptoms which rather wax and wane throughout the course of the illness.
So when we look at that, when we look at this waxing and waning of symptoms we can imagine that again, if we think about the prominence of negative symptoms relative to affective symptoms being an example of positive symptoms, the likelihood that a particular individual will make that proxy kind of diagnosis of deficit kind of presentation waxes and wanes over time as well. In other words, it’s probably fairly unreliable over time. Now, you can also think about negative symptoms in the dimensional way.
And then like we’ve, along with Greg Strauss and Brian Kirkpatrick and some of our other colleagues, Savannah Gaterissy(ph.) and so on, we’ve established a dimensional model of negative symptoms that is rather a hierarchical model with motivation and pleasure, and emotional expression being high order factors that influence five first order factors. So this was published recently within the last few years about this hierarchical five factor model, which is the model that appears to have the best validity for describing the scope of negative symptoms in a dimensional way.
I thought I would take a look at some data to kind of look at negative symptoms in older adults relative to younger age groups, looking at this in many different datasets from collaborations that I’m currently involved in, and to kind of look at the question of well is there consistency across several samples about the differences in the severity of negative symptoms, do we see some consistency across multiple different samples, and do we see this, with the proxy classification scheme for example, do we have evidence that there is a prominence of negative symptoms in older adults relative to other age groups, and then to look at some of the possible contributors to secondary negative symptoms in older adults in geriatric adults versus non-geriatric patients.
And looking at four different samples, a sample from Italy courtesy of Savannah Gaterissy, a sample from the CATIE trial, and the Spanish sample, and a sample from a collaborator with us at the Maryland Psychiatric Research Center. Looking at the Italian sample, again, these negative symptoms data are based on an Italian sample based on the BNSS, the Brief Negative Symptom Scale, and the Positive and Negative Symptoms Scale, the PANSS. You can capture the five factors that I mentioned before.
And when we look across these age groups it looks like you see that it appears to be peaking in subsample above the age of 56, and then down here you see that when you classify these by sex, men generally have negative symptom scores that are higher than those of women. And then there’s the same deal with the PANSS as well.
With the PANSS you can capture both motivational pleasure and emotional experience dimensions of negative symptoms, and we see this peaking with again sex differences where men were scoring higher, and then when we looked at the proxy classification scheme, which looks as an index of the predominance of negative symptoms, we see that, again this is also peaking with older adults, and of course in the last right corner is their level of functioning, and the level of functioning is lowest for older adults.
However, the question is is this consistent. We’re seeing this in the Italian sample, again this is cross-sectional, we’re seeing this in the Italian sample, but we’re going to look at some of the datasets and see if we continue to see that effect.
And then looking at some potential contributors to secondary negative symptoms, we see that when you compare the 56 and older age groups to the younger age groups there are differences in physical functioning, and the physical functioning scores are lower for the older adults. The other difference was in the anxiety scores. Anxiety scores was lower for the older adults compared to the other age groups.
In the basic stepwise regression, we find that for the entire sample physical functioning was one of the predictors of negative symptoms in the sample, which is quite lower for older adults. But when we look at it in the CATIE trial, we don’t see that peaking of negative symptoms in the older age groups, suggesting that this is rather inconsistent across samples.
In the CATIE trial sample we see that we’re not seeing that peaking. However, the proxy scores remain elevated in this sample despite the fact that the negative symptom scores are not highest for the 56 year and older age group. And of course, when you look at this by age and sex, you see that this is of course highest for older men than the other subsamples.
Potential contributors to negative symptoms, we see that hallucinations are much lower for the older adult sample, as well as depression scores. However, their cognitive performance is worse, their AIMS scores, suggesting potential medication effects, are much higher for the older sample, and their EPS scores are also higher in the older adult population.
And then when we look at contributors like predictors of potential negative symptoms in the sample, cognition, the depression scores, EPS scores, and their drug attitude surprisingly was also one of the predictors. And of course, the AIMS score looking at the facial was also one of the predictors selected as the optimal set of predictors of negative symptoms in this sample.
Spanish sample, again we don’t see any differences that favored the older adult population, but when you look at the very last graph here to my right you see that the proxy scores remain relatively elevated again for the 56 year and older population. There were no significant differences across the subgroups on any of the potential sources of secondary and negative symptoms in this particular sample, but across the entire sample the Calgary depression scores and the neurological evaluation score was a predictor of secondary potential contributor to secondary and negative symptoms.
And then in the Maryland sample, remarkably again, when you look at using the schedule for the deficit syndrome, which can help you identify or compute the severity of negative symptoms, as well as classify patients into a deficit versus non-deficit category, we see that the older adults actually had less severe negative symptoms in the sample.
And then we find out that in terms of the deficit classification they were less likely than some of the other age groups to have been classified as having a deficit form of schizophrenia using the schedule for the deficit syndrome, which is the gold standard for identifying deficit cases, which tells you that the likelihood of having deficit schizophrenia is no higher in older adults than the other age groups.
However, their proxy scores remain elevated. So they’re looking at potential contributors in the dataset, again Parkinsonism and dyskinesia are highest for older adults compared to the other age groups, but their hallucination scores are much lower. In terms of predictors, Parkinsonism, cognition, and social contacts or lack thereof if, you will, were some of the contributors to negative symptoms in that sample.
Now, to kind of put all of this together, almost done, although negative symptoms are highest in the Italian sample, we don’t see this replicated across samples. So this doesn’t support the contention that negative symptoms are more severe in older adults. However, we’re seeing an increased likelihood of prominent negative symptoms in each of the samples, and in the Maryland sample not related to their deficit status at all, which may suggest that perhaps what we’re looking at is an increased likelihood of proxies due to the prominence of negative symptoms, likely in relation to depressed symptoms and other symptoms.
So there are no greater odds of meeting deficit criteria in older adults, but there’s an increased likelihood consistent across samples of being classified as having a proxy for the deficit syndrome, which may -- again the prominence of the negative symptoms relevant to other symptoms. And of course, men were almost always consistency rated as having more of those symptoms.
And then of course in terms of the potential contributors, EPS, physical functioning, cognition, were most likely, as well as the orofacial symptoms most frequently contributed to secondary negative symptoms in older adults, whereas depression, anxiety, and suspiciousness contributed to secondary negative symptoms across all age groups. Now, to kind of finalize this discussion, there is no evidence that negative symptoms are higher in older adults, and the prominence, what we see is a prominence of negative symptoms.
The proxy classification scheme in the sample range from about 18 to 40 percent, just close to 40 percent in the sample. However, this again like I said before is not due to the fact that older adult patients were being diagnosed as deficit patients. This is across sectional sample, that’s a limitation.
In terms of treatment, treatments for negative symptoms in older adults should probably target not just the negative symptoms directly, as well as some of the secondary causes, like what were the cognitive deficits that may be age related, some of the medication effects are to be thought about, and the physicality of older adults has to be thought about as well, and what that does to the ability to engage in some goal directed activities.
In terms of treatment, I think that with older adults who probably think about, some of my colleagues that are prescribers, I’m not a prescriber, I think we probably think about lowering the dosing of antipsychotic medication, that’s certainly a consideration that my prescriber colleagues take into consideration, because they have fewer positive symptoms, and there’s a greater sensitivity to EPS effects.
I think with older adults it's very possible that it happens more readily when you put them on medications, we have to also think about some of the cardiovascular risk factors and other health issues that could be impacted by antipsychotic dosing as well.
In terms of psychosocial interventions, we want to think about supporting them in terms of activity scheduling, but scheduling activities while taking into consideration their physical health and their physical abilities, their mobility has to be taken into consideration, and their financial and social resources have to be taken into consideration, as well as thinking about some cognitive supports to help them walk around some of the cognitive deficits that may impact their negative symptoms as well. I’m going to end here, these are some of our colleagues and collaborators that we’ve had the good pleasure of collaborating with over the years. With this I’m going to pass it on to Dr. Iris Somner.
IRIS SOMMER: Thank you very much. I would like to focus on treatment for women with psychosis. And my plea will be that it’s time to make the difference. So I’ll start with the mix of the sex neutral brain. Women have been striving for equal rights of course, but that doesn’t make us similar to men, and especially the brain of the men is not similar, we should keep that in mind, because there are several important differences in the male and female brain that have their consequences for psychosis treatment. For example, brain metabolism is higher in women, while brain weight is higher in men. Men have 17 percent more neurons in the cortex, but the neurons of females have some eight percent more connections.
Maximal brain volume is already reached at 10.5 years by girls, and only at 14.5 by boys. Women have a higher immune response, but men have a higher stress response. Women have higher dopamine sensitivity, but men have higher dopamine production. So we’ll see how all of this becomes relevant for treatment of psychosis.
We have to keep in mind that there is a gender bias, and that what we now describe as psychosis and optimal treatment for psychosis is largely based on studies performed predominantly in men.
So I’ll give you an example of long-acting antipsychotics, which comes from a meta-analysis of Santos-Casado, and he found that while we have 132 trials investigating efficacy and safety of long-acting antipsychotics, only three of them discussed their results per sex. This is actually very important to know exactly how to dose and to time long active treatments for women.
So let’s look at diagnosis. And I’ll refer to a paper that I did using the Finnish registry, which has all men and women diagnosed with schizophrenia in Finland. And we can see that the age of onset is highly variable in men and women, men in blue, women in red.
So this is a bit different from the data that Dr. Castle presented, because he showed the onset of psychosis, this is the diagnosis of schizophrenia, and actually we see in red in women that it’s high, well, starting in the second decade, and it remains high quite up until high age. We start measuring at 65 because then the follow-up time was done, but actually it would remain high if we could follow them a bit longer still.
If we look at the symptom cluster, then we see that affective disorder are such an integral part of the symptom complex in women. So here you can see men at the left, and also in men 39 percent also have affective disorder, but this is 62 percent in women, so that’s very high. And which is well known, substance abuse is much higher in men, 34 percent, only 10 percent in women. So if we have much variable of an older age of onset, together with very predominantly effective symptomatology in women, this makes diagnosis much harder, because it’s much less the typical stereotype picture that we know schizophrenia to be.
And as a result, it’s much more difficult to recognize schizophrenia and female psychosis. And I was very happy to have this very clear interview with Dr. Roberta Payne, who really showed that it can take so long until the illness is well recognized in a woman, and that’s exactly what we see. So the duration of untreated psychosis is six years for women on average, while it’s only two years in men. And this is what happens.
So it’s not that women are not receiving any care, they are receiving care, but they’re not receiving the right diagnosis. So, if we look at the upper arrow, that’s the men, and they sometimes start with a diagnosis of anxiety disorder, then it could be other psychotic diagnoses, maybe personality disorder, substance use related disorder, mood disorder without mania and finally with mania, and then at the age of 34.5 they on average receive their final diagnosis of schizophrenia.
If we look at the lower arrow, which is for women, the order is not different, but there’s a gap. So after their last diagnosis of mood disorder with mania, there’s a gap of some six years in which these women are in care, but diagnosis of schizophrenia is simply missed. And you know long duration of untreated psychosis is quite disadvantaging. So that really worsens their prognosis.
When they are diagnosed women still receive some disadvantages because we know that specialized early psychosis care is really important, and it really improves psychosis, but since they’re diagnosed so much later, women are often too old to be allowed to early psychosis care. Furthermore, the specialized early psychosis programs are usually not that well-tailored to the needs of females.
For example, there’s a lot of information and help for drug of use, which is rarely a problem for females, while there’s no help for example with family planning, Childcare, relation management, which can be very helpful for women. Remember that these women are older, often already have families and hold jobs, so they will have quite different needs.
Then we go from diagnosis to pharmacotherapy. And it's important to recognize that both pharmacokinetics and pharmacodynamics are both highly influenced by estrogens. For example, the estrogens largely inhibit the CYP1A2 liver enzyme which metabolizes most antipsychotics. It also enhances the CYP3A4 enzyme, which metabolizes mainly quetiapine and lurasidone. Also ,renal clearance is much lower in women, and gastric emptying is much lower in women.
So if you provide equal dosing of most antipsychotics, the blood levels are likely to rise. And then if we look at pharmacodynamics, actually the female brain is much more responsive to antipsychotics. So actually, females would need lower instead of higher serum levels of antipsychotics.
So in a study we reanalyzed RCT data from the first intro study, which was done in Bergan and recently published in Lancet Schizophrenia, and we did a birth sex analysis, and we looked at blood levels in men and women, and corrected for dosing. So most antipsychotics were dosed more or les equally, they had flexible dosing, and when corrected for antipsychotic dose blood levels were much higher in women than in men for Olanzapine, 159 percent, also for Amisulpride, 172 percent, for Aripiprazole 156 percent, for Clozapine 140 percent, but not for Quetiapine.
So we saw that Quetiapine is one of the few antipsychotics being metabolized largely by CYP3A4 which is enhanced by estrogen, so indeed blood levels are more or less equal. But this is the exception. And for many antipsychotics, actually all apart from quetiapine and lurasidone, blood levels become much higher with equal dosing in the fertile age. So until menopause. So actually, for these agents, women would probably need (audio problem) direct dosing.
Prolactin is also a worrisome part. Most antipsychotics, with the exceptions of Aripiprazole and Brexpiprazole, raise prolactin. If prolactin is increased for long duration, it increases the risk for breast cancer. And the Finland group from Heidi Taipale showed that risk for breast cancer actually is increased 150 percent if women have been using prolactin raising antipsychotics for more than five years.
Now, this is worrisome, because schizophrenia women already have increased risk because of their genetic predisposition and are more likely to skip screening programs. So this is something we definitely do not want. Also, their risk for osteoporosis increases by using prolactin raising drugs, and of course menstrual egg irregularities will result.
But an important thing, raising prolactin, is estrogen, because we know estrogens to be protective for mental health in general, and for negative and cognitive symptoms in particular. So what happens if we prescribe prolactin increasing antipsychotics to women with psychosis, they will lower the estrogen levels.
Now, these are biological systems that lactating women will not become pregnant very early, so it’s a normal biological response to high prolactin levels to reduce estrogen production. However, estrogens have been so protective for these women, and they actually lose their protection by using this medication, which results in worsening of negative symptoms and deficits in cognition.
We recently finished, well, not we, actually Bodyl Brand did most of the work, and I was very lucky to supervise her. She included 156 patients with schizophrenia from whom 104 could be randomized towards 12 weeks of placebo or 12 weeks of raloxifene at 120 milligrams.
Raloxifene is a selective estrogen receptor modulator, and that means that it’s agonistic in brain and bone, so it will give you the estrogen protection that natural estrogen provides, and it will also prevent osteoporosis, but it will have antagonistic action in breast and uterus, and this means that it does not increase the risk for breast cancer like estrogen does.
And she just did the analysis, we haven’t actually published it yet, and what she found is that there was a positive effect for women only, unfortunately not for men, and we saw an improvement in negative symptoms and in cognition with selective estrogen receptor modulator.
So the final topic I want to discuss with you is menopause. So what happens in menopause? The natural estrogen production reduces. And we saw that estrogen has several effects, it inhibits the CYP1A2 enzyme, and increases action of the CYP3A4 enzyme. So this means that dose needs to be adjusted after menopause. And actually, of most antipsychotics, except for Quetiapine and Lurasidone, women will need higher doses after menopause in order to have the same efficacy.
However, women like all patients will become more susceptible for extrapyramidal symptoms and also for cardiovascular symptoms. So this is going to pose a problem, and what could be a way out is to consider Raloxifene addition, because this will also give you the same inhibition of the CYP1A2 enzyme so that you don’t need to increase those.
I’m going to wrap up. I think that we can make a small step towards personalized psychiatry in that we can give special psychosis treatment for women, especially in improving recognition to reduce duration of untreated psychosis, reduce dose of antipsychotics that are metabolized by CYP1A2, and try to avoid prolactin raising medication if at all possible, or if that is not possible maybe adding Aripiprazole, a dopamine partial agonist which will also reduce your prolactin levels. Thank you very much for being part of this excellent workshop, and I’m going to give the floor to Dr. Palaniyappan.
LENA PALANIYAPPAN: Thank you very much. Let me start by thanking all the speakers, this was a fascinating one-hour session. A lot of continuity with the lived experience that Dr. Payne was discussing earlier. So I congratulate all of you for these talks. Now we have 10 minutes of Q&A session, and then we have 20 minutes of panel discussion after this. For the first 10 minutes, may I request all the speakers to please switch on your video, but please keep your microphones muted until I call upon you. We have a few questions on the Q&A box.
For the sake of time, I’m going to give one question for each speaker to start with, and if we have more time I’ll come back and circle again. So, the very first question is for Dr. Cohen. Are symptoms of psychosis correlated with anosognosia negative and cognitive symptoms, or are they independent of each other over the lifespan of psychosis? Dr. Cohen?
CARL COHEN: Thank you. I took a quick look at some of my papers, we recently had published a paper on illness awareness. And yes indeed, increased cognitive functioning was associated with better illness awareness, and lower negative symptoms was correlated with illness awareness. Also, with negative symptoms it was a relationship with cognitive scores in another paper we did.
But the odds ratio was 0.96. In all cases it was not terribly impressive, there was some correlation, and as I showed in the slide during my presentation, there are correlations depending on whether you want to consider it low or medium, there’s something that exists, but the explained variance, or the overlapping variance is not realized between these different components.
LENA PALANIYAPPAN: Thank you. The next question is for Dr. David Castle. This is about the frequency of illness between the two genders. Is there a greater frequency for women, because there are more women in that age group. Is that a survival bias?
DAVID CASTLE: No, it is not, because what I presented were actually rates, not just numbers. And if you look at the sheer numbers and then the rates it’s very clear it’s not just longevity in females. And there was a related question if I may about any gender biases. But if you look at these data, they’re uniform pr4etty much over many years of ascertainment, so over decades in fact, and also robust to different cultures. So I don’t think you can explain it in sociocultural terms.
LENA PALANIYAPPAN: So the next question is for Dr. Anthony Ahmed. You have well wishes, Dr. Ahmed. Followed by that you have a question. How does one differentiate the diagnosis of late onset schizophrenia and dementia with psychosis. This is a perennial problem for most practitioners.
ANTHONY AHMED: It is a challenging diagnosis; it is a challenging distinction to make. I think you probably want to take into consideration, probably consider doing some cognitive assessments and incorporating some cognitive assessments, there are certain cognitive patterns that you may find in patients with schizophrenia relative to before late onset that might help with figuring out the odds. Again, you’re playing the odds here with the level of cognitive assessments because they’re not diagnostic in any kind of way. It’s like looking at a broken bone.
But you can certainly use some cognitive data to inform that. There are certain areas of cognition that are more crystalized, and certain areas of cognition that are more fluid, and you could potentially take a look at these patterns of cognitive scores to try to inform along with clinical ratings and get a lot of collateral information that can have premorbid adjustment over time.
With older adults there may be fewer people that can speak to their premorbid adjustment unfortunately, because at that stage in life you are looking at people that have been, family members that may have passed away, so there’s fewer collateral information available to help ascertain that diagnosis, and that’s also part of the challenge as well. But if that information is available, you can pull together a lot of this collateral information, as well as clinical ratings and cognitive data, to try to assist with making that diagnosis.
LENA PALANIYAPPAN: There is a question for Dr. Iris Somner, the next question. This is about frequency of trauma in women. Is there a relationship between higher frequency of trauma and frequency of psychosis in women?
IRIS SOMMER: Yes, definitely. Unfortunately, trauma is even more frequent in women with schizophrenia, especially sexual assault, and several studies have found an association with age of onset. We could not replicate that in our sample, but this has been found, and there have also been reports on higher levels of negative symptoms in traumatized women. We could also find that in our sample.
LENA PALANIYAPPAN: Thank you for that. We have time for a couple more questions. I’ll circle back to some questions at the top of the list. This is for Dr. David Castle again. Evidence for efficacy of estrogen therapy for treatment of schizophrenia is weak. What is your experience?
DAVID CASTLE: It is weak, but there is some. Joshua Kilcarney(ph.) in Melbourne has done some really nice work showing some effect in acute schizophrenia. And they’re quite nonspecific actually. So, my bet is that it’s not necessarily antipsychotic as such. And she showed, unlike Iris’s SERM study, efficacy for men as well, that has been replicated, but the more exciting stuff I think is SERMs because they’re safe, Iris just presented some data, and there’s other work from Tom Weichert and others in Sydney and Australia showing the same sort of thing, effects on cognition.
I’m not sure about the effects on negative symptoms. I’m always worried about negative symptoms, because so many of them, as Anthony pointed out, are secondary to other things, and you certainly get morphic effects of estrogens and SERMS on mood and other things, which might impact. So that’s a bit of a long-winded answer.
LENA PALANIYAPPAN: Thank you. If you don’t mind, Iris, there’s one more question, maybe we can finish the Q&A session with a last question to you, Iris. Is the later diagnosis of schizophrenia in women, is this related to gender bias, is there any true difference in the onset of illness?
IRIS SOMMER: I think it is both. I think one thing is that there is the protective influence of the estrogen, so it will be later. The other thing is the picture really is much less typical than in men, or actually perhaps we should learn to better recognize the female type of psychosis.
I think that in the Netherlands they’re often received, for example a diagnosis of borderline personality disorder, or psychotic depression, and then it will take several years until they receive the right diagnosis, and also the right treatment. So perhaps there is let’s say a development of symptoms, so there really was more like that, but if you don’t have the negative symptoms that prominent it will be more difficult to recognize it.
LENA PALANIYAPPAN: Thank you very much for all of your patient answers. And thank the audience also for the questions. Unfortunately, we couldn’t go through all the questions, so may I please request the speakers to look at the Q&A box and engage with the audience who are asking specific questions. We will now move to the discussion session. Can I request all the speakers to just mute your mic now, but please stay online, on video, and I will request Dr. Frangou and Dr. Lee to come to the panel.
SOPHIA FRANGOU: Thank you very much. Thank you to all the speakers, these were amazing presentations. I think the point of this discussion panel is for all of us to engage each other in the discussion and reflect both about the things that we’ve heard and the things that perhaps we need to do in the future, focusing more on future directions. I did have a couple of questions that are to some extent for all the speakers, and would it be okay Ellen if I start?
ELLEN LEE: Absolutely.
SOPHIA FRANGOU: Thanks. I think a lot of the time people discuss the issue of heterogeneity in many sort of forms, heterogeneity of risk factors, heterogeneity of presentation, severity of symptoms and so on, and of course I may have missed it entirely, I couldn’t see any attempt perhaps, or maybe it’s in other studies that were not presented here, to approach this not in a variable based sort of methods, like which variables are associated with which variable, but on a person based sort of basis, looking at clusters, and perhaps we can identify more specific constellations of people, men and women that have more discrete profiles that can explain why some things do not replicate across studies.
And to understand a little bit the relationship between sample composition and clustering that will help us understand these things. So I wasn’t sure whether these sort of considerations have been part of discussions with the groups or the presenters that we had today.
DAVID CASTLE: I think it’s a great question. So you are absolutely right. We should do more of that. And now we have the abilities to network metanalytic sort of techniques and actually do that, and establishing nodes and how things interact. We did a very much less sophisticated analysis with the camel register sample with Pat Schaum(ph.) leading us, which was led in cross-analysis, and we focused on sex and age at onset, on really nice clustering. This dementia praecox group really comes out separately, the later onset paranoid group or paraphrenia according to Krapelin was a more affective type group. But I think you’re right, the sort of network analyses and stuff we should definitely be looking at.
ANTHONY AHMED: Can I speak to that? In some of the data that we’ve published we’ve used latent class analysis like Dr. Castle was referencing, and some of these other latent variable mixture models. There are some models, competitional models that are available for looking not just cross-sectionally but even looking at trajectories.
The method is called latent growth mixture modeling, that if you have change data over time you can look at trajectories, so that you’re able to parse heterogeneity not just in terms of what people are like at this timepoint one, but how people change over time, if there are subgroups of people. And then subgroups of people, they decline, they improve and so on.
And then you can also as part of that modeling you can also look at well what are some of the characteristics, you can probe those subgroups of change, or those mixtures of change over time, you can look at what’s different about this subgroup that’s improving over time versus this subgroup that’s declining over time, or this subgroup that remains relatively stable.
And these methods are out there. I think the challenge is that we often don’t have enough training to be able to leverage the sort of technologies with a lot of data sets that are out there. But that kind of quantitative modeling has really advanced, and there are lots of programs that can help with this sort of analysis, including latent class analysis being one of those.
CARL COHEN: I was just going to indicate that one of the problems is the outcome measures that are used. As I pointed out in one slide, there are 120 different combinations when we use five outcome measures, and these things fluctuate over time. So I guess you can begin to find some of this doing sophisticated analyses, but it’s very complicated given the combinations and the fluctuation in the different outcome measures. So, it’s quite daunting. I guess it can be done, but it can require huge sample sizes.
SOPHIA FRANGOU: I think since the NIH people are on the call and are listening, we could sort of say that in the same way they have supported first episode psychosis, prodromal psychosis with big networks, data harmonization approach, perhaps what we’re highlighting as a need in terms of understanding, other people getting older with schizophrenia, or older people getting schizophrenia, which are two groups, we need some sort of perhaps network and some sort of coordinated approach towards harmonization of measures and so on, which has been done as part of other networks. Just a thought since it’s been recorded.
LENA PALANIYAPPAN: This reminds me of a broader question that probably we don’t have an answer for it, but integrating all these talks, the gender issue, the interaction within sex and age that was brought up, one question that comes up is many chronic disorders have differences in age-based presentation between the two genders.
Many males get more unwell with many chronic disorders at younger age, but after menopause there is some leveling that happens for many cardiovascular brain diseases. But we do not treat those disorders as two different diagnoses, two different antagonists in most cases. Should we continue to treat late onset illnesses different from early onset illnesses? What are the merits and demerits of it? It would be interesting to discuss that.
DAVID CASTLE: I think Iris talked about this. Nuancing the choice of antipsychotic and the dosing. Mary Seeman actually in Toronto did a lot of very important early work on that, so looking at that. I also think the psychosocial wraparound, so the later onset patients who I remember really striking me about this came into our early psychosis service, most were young men, and she was a woman in her 40s with a career and three kids, and her needs in a psychosocial sense were completely different from all the young men. I thought our services aren’t setup for this very well. That’s why I’m so excited about this workshop because it actually says they’re also first episode, they also deserve all the bells and whistles which we give to the younger people.
CARL COHEN: Could I ask a question about the gender differences, or raise some points about it? It’s true women certainly develop schizophrenia early. More women develop schizophrenia earlier than in late life. We focus on the issues of estrogen, but if you look at absolute numbers it’s much greater chance for a woman to develop schizophrenia when she’s younger.
The differences between genders change, but the absolute numbers are higher. I’d also raise the possibility though, I don’t have any mechanism for it, are men protected in some way as they get older, is there something about males that protect them from developing schizophrenia as they age. So I raise that as another point.
DAVID CASTLE: There is a whole debate about this, Carl you would be aware, where it’s typologies or whether it’s estrogen effects. Estrogen effects I think are overstated personally, but that’s my view. The very late onsets have nothing to do with estrogen, right? But one thing which the male brain might be somewhat protected with, of course it’s a simpler brain, it might be bigger but it’s simpler as Iris has pointed out, dopamine receptors in males start at a higher level, and then come down like this, whereas females tend to hang on to their dopamine receptors. And I always wondered whether that had something to do with this, and also said something about why older women or not so young women are more vulnerable to auto dyskinesia as well. Pure hypothesis.
IRIS SOMMER: I tend to say that also drug abuse is a factor in younger men and not so much in older men, which is not really protecting but also not provocative for older men.
LENA PALANIYAPPAN: This is a fascinating discussion. One other point that comes up is the utility of studying phenomenology, why do we need to study the phenomenon and the symptoms. I think one of the very nice developments in the last couple of decades in early onset psychosis is the ability to develop paradigms around who is at risk and trying to follow people up. Is this time for us to think about the at-risk symptomatology for late onset illnesses? If we’re going to think about it, what are the starting points? It would be interesting to discuss that. Any thoughts?
IRIS SOMNMER: I will concentrate on the women, because duration of untreated psychosis is so high in this gender. I think we need to do some screening in affective disorder programs for psychotic symptoms in women, and I think we might find quite some missed schizophrenia diagnosis there.
DAVID CASTLE: The other thing I would say is we’ve got to remind ourselves that the clinical high risk is interesting but very inexact. If you look at probably the best studies in the field, the conversion rate is relatively low actually. People are saying up to 30 percent over three or five years, but in most hands it’s not that good.
And the other thing is, remember, the major risk factor is actually health seeking. So if you look at the sheer epidemiology, Jim Fornos(ph.) keeps talking about this, if you take a general population sample and you look for these high risk features you’re going to find zero by way of predictive validity. If you take a selected health seeking sample, it’s a completely different story. And we know from Chris Hollister’s work in the UK 30 years ago, just coming to a psychiatrist is going to predict something, you’re distressed in some way. And people don’t seem to factor this into their modeling very well.
LENA PALANIYAPPAN: It’s very interesting. Again, gender plays a role in health seeking behaviors as Iris pointed out.
ANTHONY AHMED: There are some changes in the environment as well. The environment in which the illness happens is dynamic across the lifespan, even when you live in the same neighborhood there are some changes that happen in your life. The ability to model that in a way that’s consistent a very methodological challenge, and the ability to quantify that, research for a longitudinal research study would be interesting to see if somebody takes that on, and how they do that.
CARL COHEN: Can I raise another question about phenotypes? I thought it’s important for us to discuss this to lay some groundwork for future talks. Are we satisfied with the cutoff for late onset disorder and very late onset disorder with respect to age? I mean there are very few diagnoses in medicine that are based on age, and yet we have that here.
The other thing is when we’re talking about older adults with schizophrenia, are we happy with the age cutoff for that? Should it be 50, 55, we use 55 because we feel that there is an accelerated aging of sorts in people with schizophrenia. There should be some more or less consensus on it, we’re going to do research on this, even if we’re wrong initially, we’re going to have so much mixed data. When I review literature it’s always these issues about inconsistencies and definitions of what we’re talking about.
SOPHIA FRANGOU: Thank you about this point. When you were talking initially I thought that you raised a very important issue, when does old age begin, full stop, which doesn’t have in itself a definition that I know anywhere, it’s kind of historical and based essentially on occupational status, working age, non-working age, that’s the most common - but again, coming to my sort of original discussion, I don’t think we can be totally conceptual about it, because we don’t have a good guide as to how do we theorize about the modelings.
I think this is another opportunity for data driven definitions, looking at different clusters and saying this cluster is very different from everything else, and the typical onset of this cluster is 65. So I would call that, whatever. That’s my take anyway.
CARL COHEN: The definition of nonaffective psychosis, I don’t understand that definition. We see schizoaffective disorders, I have plenty of patients who have very high rates of depression, schizophrenia patients. What does that definition mean? Are we going to use that also as a criterion for our studies? I don’t know.
ELLEN LEE: That is a really good point, Carl. I think when we conceptualized this workshop, we were trying to separate out sort of schizophrenia spectrum disorders versus bipolar disorder or depression with psychosis, just because they have different courses, and we want to be more refined, but I can understand the confusion, because you’re right, one of the greatest indicators of remission or recovery is related to mood symptoms, and it’s an important thing to consider. So I think that’s one of the things that might be imprecise in how we speak about this, and it’s much more heterogeneous and confusing.
I wanted to see if we have time to pivot into how to overcome some of these barriers, how can we recruit diverse populations. I think many of you have shown that having different samples from different groups showed different trajectories across time. And so I was wondering how you felt like we should overcome this, other than building a giant consortium as Sophia has proposed we do.
DAVID CASTLE: One thing I would say is we have to restrict our exposure to DSM. If everything gets defined in terms of DSM, that’s going to be the lens. That’s not right in my view, and we’ve got to get back to proper phenomenology and understanding epidemiology and epidemiological context, and that can inform age at onset discussions, and then phenomena, it’s about phenomena and about people, it’s not about DSM, with all due respect.
LENA PALANIYAPPAN: That’s a great point. One more thing that we sometimes forget is a lot of research activities follow how services are structured. Increasingly with early psychosis being separated from the midlife and late psychosis, research activities are also getting insular. I think this is really important that we think about lifespan perspective across the disorder. Any more comments on this session? We have two more minutes.
IRIS SOMMER: I much applaud the initiative of Dr. Cohen to skip the nonaffective part of psychosis, I think it will really help to better recognize psychosis, as there is often a negative component, a negative mood component, also in men, not only in women.
LENA PALANIYAPPAN: There is a show of hands, Dr. Carol Tamminga.
CAROL TAMMINGA: I just want to say that I agree with you entirely that everything has gotten divided up into little groups, not only has it gotten divided up into little groups, but the divisions are not based on data, the divisions are just based on, I don’t want to say fantasy, I do want to say that they’re probably based on our conceptualizations, and our conceptualizations aren’t necessarily based on data.
So I do think that it makes some sense, at least it has been helpful to us, to go back to a dimensional diagnosis like psychosis and see what are the characteristics that fall out in a data driven way. It’s been enlightening to all of us actually. We seem to be on the verge of doing something that is going to be databased and pull the field together, so I’m looking forward to it. Something like this meeting here that we’re all gathered together to talk about.
LENA PALANIYAPPAN: Thank you Dr. Tamminga. That is a very exciting example to close this session. All the speakers have been very punctual. I don’t want to be the only unpunctual person, so I’m going to call this to a close. Can I also remind all the audience that there’s going to be a 15-minute break from now, please come back in 15 minutes. And thank you again for the organizers and the discussion panelists as well as speakers. This was an entertaining session, thank you very much.
MICHAEL GREEN: Hi, I would like to welcome people back for the second session, which is going to be on cognition. I am Michael Green. I'll have the privilege of moderating this session.
The first speaker is Dr. Roman Kotov, from Stony Brook University. Then Dr. Eva Velthorst from the Icahn School of Medicine at Mt. Sinai, Dr. Lawrence Yang, from New York University. And as the sole presenter who is not from New York, Dr. Elena Ivleva, from the University of Texas Southwestern Medical School.
I'll remind the speakers that they should keep to 15 minutes for presentation. At this point I'll turn it over to Dr. Roman Kotov to start his presentation.
ROMAN KOTOV: Let's begin. So, I will tell you today about excess prevalence of dementia in schizophrenia. I will elaborate this with some findings from Suffolk County Mental Health Project, and propose my ideas for what may be going on.
Starting with the first item, we have thought for many years that schizophrenia is a neurodevelopmental disorder. Cognition declines early on and then stabilizes about the time of first onset of psychosis and stays stable, maybe improves. But these studies are based on short follow-ups and in relatively young participants.
So new data that's coming in, turning this wisdom on its head. For example, this is a study of Medicare beneficiaries, where looked at prevalence of dementia and in those with schizophrenia, at age 66 it was already 28 percent. And at age 80, it reached 70 percent, which is vastly higher than rates of dementia in the general population, as you can see here in dashed lines.
So this is actually quite staggering, and other data consistent with it, for example, this is dementia as cause of death, in a population-wide study in the UK, and they compared the causes of death in those with serious mental illness versus those without, and no deaths under age 45 were attributable to dementia. But after age 45, rates were five to ten times higher of death due to dementia in serious mental illness.
Altogether, there are more than 15 registry studies, and they're all in agreement that there is a dramatic excess of dementia. However, these studies, started in older participants, those who are aged 65, or at least 50 or older, and it is missing those with schizophrenia who recovered from the illness and hence are not part of healthcare system, or those who haven't recovered but left, and certainly those who died; quite a number have.
And so another problem is administrative diagnosis. These are diagnoses of clinicians indicated in the medical records, and they are often based on the first meeting with the patient, and in schizophrenia and dementia have many features in common, shown here in purple, and it is easy to confuse impairments due to schizophrenia to those due to dementia, especially if you don't have a long-term baseline.
But it also can go -- so it would inflate the association, but it could go the other way around as well in that we can have basically the physicians not assigning diagnosis of schizophrenia, or dementia to those with schizophrenia, assuming that impairments are due to schizophrenia when in fact they are new and escalating.
Also, of course, from administrative data, we cannot learn about trajectory of decline, when it begins, how quickly progresses, and neuropathology typically is not part of records.
So here, Suffolk County Mental Health Project may be able to help. This is a countywide epidemiological study in Suffolk County, New York, that included first admission patients in all 12 inpatient units of the county; 72 percent of them ultimately participated in the project, making this one of the very few representative studies in North America. It was designed, launched by Evelyn Bromet. Evelyn remains very much involved. Six hundred and twenty-eight participants have been recruited by year 25. Mortality of course took its toll, but we still maintain 70 percent retention in the survivors. And this also makes it the longest follow-up of first-episode sample to date.
At the year 20, we recruited neuropsychiatric comparison group. We call them neighbors, because they are geographically as well as demographically matched to cases.
And this the key slide. If you remember any part of this presentation, this is the slide. We have access to school records of study participants, so we are able to -- and then of course we have done our own neuropsychological testing, so we were able to map a course of cognition, at least general intelligence, from basically age 17 to 77. And the blue line here is participants with schizophrenia, and as you can see that as children they had normal intelligence, IQ score of 100, but about 14 years before onset of psychosis, their cognition began to decline, and decline continued, contrary to expectation, unabated. If anything, it accelerated, 22 years after psychosis onset.
What this finds, 30 years at least post-onset, the average participant with schizophrenia lost 18 IQ points. If this trajectory continues, we will see the same rates of dementia as epidemiology of all the participants suggests.
But of course, dementia is not just cognitive change. It also community functioning. Fortunately, we were able, we had the data to rate community functioning for the same period, and we use GAF scale here on the y-axis, numbers expressed on GAF scale. As children with schizophrenia had GAFs in the lower 60s, basically mildly impaired, but several years before onset of psychosis they had begun to decline, and this decline has been unrelenting, and at year 30 post-onset, the average GAF score is 35. Quite a dramatic change from the low 60s to 35.
But more directly, at year 25 of the assessed study participants, using clinical dementia rating scale, and interviewers doing these ratings, took into account functioning of our participants at baseline, that is already onset of psychiatric disorder, both cognitive and community functioning, and the rate it changes. And also, I want to spotlight that average age of the sample at this point, is 53 years of age. So quite young to have dementia. As you can see, neighbors, these are cross-sectional scores, by the way; the neighbors, the rate of dementia is quite low as expected. In schizophrenia, 12 percent declined enough for this decline on CDR to be in the dementia territory, and 32 percent declined partways, consistent with mild cognitive impairment, and they are at risk to continue the decline.
Also, these CDR change scores show expected correlations that you'd expect of a good measure of dementia, such as of course activities of daily living, diabetes, of course a major risk factor for dementia, et cetera. One correlate that is not there is age. And to me this suggests that we are not looking at normal age-related accumulation of cognitive problems, but a different process.
An interesting factor here is the association with exposure to antipsychotics. To illustrate this a little more, let me show you details of this data that we were able to use study records to reconstruct antipsychotic exposure across the first 20 years of the study. And just for illustration, we did median split and plotted trajectories of those with long exposure versus shorter, and those with longer exposure declined faster, resulting in by year 30 a difference of 6 IQ points.
We didn't do all that much with neuropathology as of yet, but we were fortunate to do a small neuroimaging study at year 20, and we found that 24 percent of participants with schizophrenia had lacunas indicating microstrokes. Only 3 percent of neighbors did.
Also cortical thinning, overall, that's not a surprise. Cortical thinning is present even earlier in schizophrenia. But here the location is interesting; it's particularly concentrated in the prefrontal cortex, and it is associated with CDR ratings of poor memory function with antipsychotic exposure, suggesting that it might be getting at the pathophysiology of this form of dementia, perhaps.
Three ideas for what may be going on. One is schizophrenia is sometimes hypothesized to be a disorder of accelerated aging, and we will hear quite a bit more about this tomorrow, I believe. If that's the case, if that's what explains the excess of dementia, then basically people with schizophrenia biologically are older, by a couple of decades, than you would expect, then Alzheimer's would be the most common form of dementia.
Postmortem studies done so far have not found Alzheimer's pathology in schizophrenia, but postmortem samples tend to be small and quite unrepresentative of people with schizophrenia. Now that it became available only in the last few years, we have reliable blood-based biomarkers of this pathology, such as amyloid neurofilament light chain, tau protein, and so we can study this question on large scale and representative samples, and we can use these markers to predict future cognitive decline. So that's the work that is yet to be done on a real scale.
Cerebrovascular diseases is a second possibility. In fact, in schizophrenia, you have hematologic evidence of a high rate of risk factors for cerebrovascular disease such as metabolic syndrome, low physical activity, and also twofold increase in diagnosis of cerebrovascular disease, but these are administrative datasets, by and large, so what they're analyzing are typically major strokes, things that come to clinical attention.
In fact, covert cardiovascular disease that is not clinically detected can also be affecting cognitive functioning, and we have reliable radiologic markers of it, such as white matter hyperintensities, lacunas and so on, but we don't have epidemiology on distribution of these abnormalities in schizophrenia, especially in older age.
So we don't know how common cerebrovascular disease is. In fact, maybe the whole story behind cognitive decline. We don't know. A concerning possibility that this may be an iatrogenic effect of antipsychotics. So first of all, exposure to antipsychotics, degree of exposure, predicts gray matter loss in people with schizophrenia. One randomized clinical trial found that antipsychotics actually cause this loss. Gray matter loss, a degree of gray matter loss, correlates with a degree of cognitive decline. Antipsychotics recently were found to predict dementia onset in people who received antipsychotics before their dementia diagnosis. And also that two randomized clinical trials that found the discontinuation or reduction of the antipsychotics improved cognition.
There is also a biological possibility, a pathway that may lead from chronic exposure to antipsychotics to impairment in prefrontal cortex, resulting in dementia. But these are just hypotheses. There are many things we don't know, and because of methodological problems I mentioned earlier, we don't have precise trustworthy estimate of incidence of dementia in schizophrenia. We certainly don't know what type of dementia this is. We have hypotheses about risk factors and pathways, but they remain to be tested.
It seems to me that the kind of design that we need is definitely representative sampling and ideally prospective studies, but if we cannot wait for decades to see the end of the study, for dementia, we can use follow-back design, where we identify a cohort retrospectively prospectively and then try to find these participants to study them now. Whichever way we go, it's very important to get cognitive and functional baseline to be able to detect this change and use expert diagnosticians who can separate problems related to dementia from those of schizophrenia.
We do need data on risk factors and antipsychotic exposures with high precision and across the life course. And we also need data on neuropathology. Full disclosure, I'm of course quite biased in thinking that this is the design, being an epidemiologist myself, but these are certainly problems that I encountered in the literature at this point.
So to the perhaps conceptual question, is schizophrenia a neurodevelopmental disorder or a neurodegenerative disorder? Well, this data seems to suggest that it is both.
I want to thank many people who have contributed to the Suffolk County project over the years, making this the landmark study that it is.
Let me turn this over to Eva Velthorst.
EVA VELTHORST: Thank you, Roman. Thank you also for touching on the Suffolk County Mental Health Project, because I will be sharing some of your data as well.
Over the next 15 minutes I will present some of the available evidence on the lifespan trajectories of social functioning and of social cognitive functioning in individuals that are diagnosed with a nonaffective psychotic disorder.
I will start with a brief overview of where it starts and the severity of these deficits, after which I will go onto the main topic of this presentation, namely what happens after first diagnosis, and then finishing off with the many unresolved questions that still remain.
We know that within schizophrenia studies, the premorbid social functioning has mostly been measured in retrospect. People that already have a diagnosis of schizophrenia and other nonaffective psychoses are generally being asked how well they functioned premorbidly, or their relatives are being asked. And one of these early studies is a study by Heinz Hafner and colleagues that showed that in the six years prior to a first admission for schizophrenia, the percentage of those being in a stable partnership or being married significantly decreased.
So longitudinal clinical studies are very rare. Obviously, they need very long follow-up of large groups of children to see how social patterns diverge and why some develop schizophrenia while others do not. There's an exception in national registry data that can be used for this, such as the Israeli National Draft Board, where nearly all Israeli males at the age of 16 or 17 are being tested for eligibility in the army. Part of their eligibility assessment is a very largescale community functioning battery, and their data is then being linked to later hospitalization registers. From this we can clearly see that among the three key community functioning deficits, people who will later develop schizophrenia already report social functioning deficits long prior to the illness.
So when we can zoom in on this particular domain and we measure social functioning really to ask about friendships and how well people have social contexts, you could see in the red line annotations that first deficits are already noticeable up to, in individuals that were being tested, 15 years prior to illness onset. But you could see also that there's a steep decline in those individuals that were being tested from 5 years before illness onset to the time of illness onset.
We don't really know what causes this immense drop in social functioning prior to illness onset, but what we do know is that social cognitive deficits are one of the main causing contributors to poor social functioning in schizophrenia. So the field is still relatively new, but what we know, this is a meta-analysis by Savla and colleagues, that social cognitive impairments are noticeable across domains. And most prominently, we can find deficits in mentalizing, the domain of social perception, and emotion perception. Emotion perception can be like recognition of faces, really, to make sense of the world around you.
For social cognition, we don't yet know when these impairments start exactly, but from clinical high risk studies, studies to clinical high risk populations that at risk for developing a psychosis, we can already see deficits across domains that are around half the size of those within individuals with schizophrenia and other nonaffective psychosis diagnoses.
So what can we take away from this first part? We can see the premorbid deficits in social functioning but also in social cognition are probably noticeable across domains, and specifically social functioning deficits, we know of those that they decline steeply prior to and around the first psychosis onset, and we still need to investigate more when the exact onset is of social cognitive deficits.
Moving onto the trajectories after a first diagnosis. I will skip this part a little bit because Roman also already explained the Suffolk County project, but we specifically looked at data of social functioning. To see what is going on with the social functions after a first diagnosis, we also made use of this wonderful data where people, 485 of the study participants, were also being assessed with an elaborate social function assessment. So where people were asked about the level of participation in social activities, but also how well they were able to keep and maintain their friendships. So these people were then again seen at six months, two years, four years, 10 years, and 20 years follow-up, when also there was this never-psychotic comparison neighbor group that was being seen as sort of a benchmark of how well people function after 20 years in the disease.
We applied a latent class analysis, as David Castle mentioned before, where we tried to cluster individuals into different functional groups, and what we see here is that we found four very stable trajectories over the course of 20 years into the illness. For all of these clusters, we see that in a few years after a first psychosis, we find some sort of bouncing back from the initial more severe social functional impairments, after which there seems to be a very stable trajectory. Here in black, you can see the comparison group as sort of the benchmark, as the relatively preserved social functioning group.
I need to note that among the profoundly impaired class, these are really most people with schizophrenia, fell relatively more often in this particular group, in comparison to those with a bipolar disorder with psychotic features or a major depression with psychotic features.
We also looked at predictors. We did not find many, apart from negative symptoms at baseline, with the biggest predictor really was their level of social functioning before they actually developed the illness onset.
Again, very long follow-up studies of social cognition have yet to be conducted. From Dr. Green's work, lab work, we know that from their cross-sectional studies that social cognitive deficits seem to be fairly stable across illness phases. So this is a cross-sectional study where they looked at social cognitive functioning across three social cognitive domains, and as you can see, in the blue line, that specifically after correcting for multiple comparisons, they did not find any significant differences across these illness phases.
So that group, this time led by Dr. McCleery, also looked at this longitudinal, so they followed the same individuals over a five-year period, and here they found an extremely high correlation of around 0.7, between their baseline social cognitive scores and social cognitive scores at five-year follow-up. So we need to note that these studies and all these studies really that have been done, most studies that have been done to social cognition, have included mainly fairly young individuals. I think the average age of individuals, even in the chronic stage, in these studies, were around 35 years old. So the main topic of this presentation is obviously what happens in midlife and after? And we don't really know that.
So for the purpose of this presentation, I ran a brief pilot study using data from the Dutch group study, it's a genetic risk and outcome of psychosis study, that followed nearly 1,000 -- 993 -- individuals with a psychotic disorder, 1,219 relatives, and 543 controls. And all of these individuals were also being assessed with two social cognitive tasks. One mentalizing task, this a hinting task, and the degraded faces task. So, the emotion recognition task.
This is just their study characteristics. You can see that the relatives here function in between the controls and the patient group, and we just wanted to investigate whether there was an age effect.
First, looking at the mentalizing task, I treated age continuously here, this is just for visual purposes. Even though it appears that individuals that are a bit older seem to function a bit better in the mentalizing task, there really is no significant difference across ages, especially when you control for sex, for illness duration, or for IQ scores.
A little different picture -- when you want to see what's really going on, you obviously need to also look at the controls and at the relatives, because they might suddenly improve way more than the patient group. But we actually find a very static deficit. Here you can see the relatives in red and the control groups in blue, and across ages, mentalizing processing skills seem to be fairly stable across ages.
You see a little bit of a different picture when we look at emotion perception tasks. Overall, this is within the patient group, we see that the older one is, the worse they get on this particular task, which is interesting. Also after adjusting for sex, illness duration, and IQ, which is interesting, because if you would think that the shorter illness duration might be protective in this regard, it seems actually that the older people that also had a shorter illness duration within the study still showed worse results. Again, we looked at the comparison between also the relatives and the control samples, and we again find no significant age group intervention.
Zooming in on a different emotion, interestingly, it seems to be that across ages, patients seem to be fairly stable in their ability to recognize neutral faces, but it seems -- they seem mostly getting worse in the recognition of fearful faces, a little bit in angry faces, and also somewhat in the recognition of happy faces. Again, no group-age interaction. So this pattern is just similar for each of the groups.
So what do we take away from this? I think we can say with fair certainty that without intervention, social impairments and also social cognitive impairments appear mostly static across the ages and across the illness, up to age 50 at least. If there is any age-related decline in social cognitive performance, it does not seem to exceed that of their relatives or of the general population. But we do know, in another study, and also in what Roman just said, that for nonsocial cognition there is tentative evidence for a quicker decline in some areas, and we still don't know whether this holds for social cognition. So there's many other unresolved questions.
Again, all the studies so far have, to social functioning, to social cognitive functioning, do not really go that much beyond midlife, and there may be a second period of decline, specifically if you look at the studies by Dr. Harvey and colleagues, which suggest that at least in functional domains, there may be a second period of decline beyond the age of 60. It would be interesting to investigate it for social functioning in particular, as well, as in the general population, chronic social isolation has been associated with poor health, and it might be that specifically at the later age, the impact of social isolation could contribute to potentially the poor health in individuals diagnosed with schizophrenia.
Again, we need studies with follow-up controls, as Roman also suggested, or follow-back controls, that are needed to determine the actual level of decline. And finally, we need to consider, as was already touched upon as well, the heterogeneity between individuals, sort of provide handles, what causes someone to be resilient and others not? How specific are social cognitive deficits for nonaffective psychosis? This is an open question. I think the literature thus far is very mixed.
Specifically, when you look into social functions and social cognitive functions, and the tests, we need to consider cultural differences, and social functions, social impairments, might mean something really different in different cultures. And thus far we have not been able to really trace the biggest predictors of decline.
That leads us to what do we need to -- in the next step forward, we need to focus more on mechanisms that underlie social cognitive impairments and social impairments, particularly in the social cognitive decline and social decline. We know there are several brain networks that are related to social cognitive impairments. We don't know yet whether these same networks are also involved in social cognitive decline. That's worth investigating as well as I think there's currently more and more new measurement techniques that really get closer to social processes as it occurs, in daily life, such as experience sampling or measuring of the brain activity during real-life conversations, and I think this could potentially increase our knowledge.
Thank you very much.
With this, I would like to hand it over to Larry, and thank you so much.
LAWRENCE YANG: Thank you for that gracious introduction, and I'll get going. The title of my talk, Association Between the Duration of Untreated Psychosis and Cognitive Performance in Older Individuals with Chronic Untreated Schizophrenia in Rural China. We've been fortunate enough to be supported by two R01s via Fogarty and NIMH, and I am the grateful recipient of the Maltz Prize for Innovative and Promising Schizophrenia Research, based on this project.
Thanks so much to the first two speakers for really setting the stage. I don't have to say much about this. Of course, we know neurocognitive impairment prominent in chronic and first-episode schizophrenia. It's among the specific domains, the other speakers have touched on them: attention, memory, language, motor, and executive dysfunction. Medium to large impairments in treated first-episode psychosis, which are linked to real-world function.
This slide is quite busy, but let me please direct your attention to the bottom right-hand side, and this is -- Roman talked so nicely about this already -- just in brief, basically normals are characterized by the dotted line. Regrettably there's a slight, modest cognitive decline as we age. People with schizophrenia, though, start off lower, there's a bigger drop around the clinical high-risk period, and the conventional wisdom went that there was the lion's share of the cognitive decline took place at first-episode psychosis, followed by relative stability. However, as Roman so nicely presented, this conventional wisdom has and continues to be challenged by new and emerging findings.
To separate -- we have another opportunity with this project to come at this question from another vantage point. To separate effects of illness versus treatment, other speakers have raised the potential effects of antipsychotic medications; we need to examine cognition in completely untreated psychosis, in an ideal world.
So, it's not conclusive yet, whether there's ongoing decline in cognition with long-term treated psychosis, but if such cognition decline is occurring, is it due to disease, antipsychotic medications, or both? So we have this project examining cognition of untreated psychosis, and I'll introduce you to my study team in just a little bit. It is what we believe to be a historic opportunity to examine the natural history of untreated psychosis in the community, and to compare this with a matched treated sample of individuals with psychosis and community normals.
As you can see below, we have a target of 400 completely untreated individuals with psychosis, 400 matched treated individuals with psychosis, and then 400 community normals. We're going to hit about 300 in each of these groups, it looks like, due to COVID-related delays. And today I'm going to be focusing specifically upon the untreated group, because this was our first paper and set of analyses, and with the understanding that we have more work to do in comparing them with the matched groups.
Another important mote, we are able to do this study in an ethical way because China has the largest treatment program in history, where they are identifying all the new people with psychosis are being identified, whether young or old. They receive free medication treatment. So after we evaluate them, we then refer them on to treatment.
Our study team, I am delighted -- our study team is a group of all-stars, with of course the exception of gender diversity, which we recognize, and hopefully, I would like for us to address over time. Currently we have Dr. Michael Phillips, perhaps the preeminent psychiatric epidemiologist in China, who really runs everything on the China side. We have Dr. Jeff Lieberman, who is chair of psychiatry at Columbia, Dr. Matcheri Keshavan, chair of psychiatry at Harvard, Dr. Ezra Susser is a renowned epidemiologist at Columbia, and Dr. Bill Stone is our neuropsychology expert.
The main aim of the study is to examine overall effects of antipsychotic medication on cognition, and due to modestly beneficial antipsychotic effects overall on schizophrenia, we propose that cognition will be worse in untreated individuals with psychosis. We are working in the remote region of Ningxia, which is a western province in China. You can see in this adjacent slide, Yinchuan is the capital of the province, but we work in the southern part of the province because this is where most of the untreated folks are.
Let me give you a snapshot. This is not your Beijing, Shanghai China. This is really remote interior China. Some of the folks that we have worked with still live in these traditional cave dwellings which I have depicted here. This foreboding figure at the top there is actually the patient's mother.
We meet the patient, who is here in green. She actually comes, this particular patient is actually pretty well functioning. She's coming in, she's working the fields. She has two kids. And she greets us kind of quizzically, saying what are you people doing here? And I think, oh my gosh, we came all this way, 30 hours of travel, in order to find this person, and she's going to screen out. But then as we assess her, it turns out, the past 12 years, she's been hearing the angels of the universe tell her the secrets of the world, so she meets criteria for schizophrenia pretty clearly.
We then verify untreated status. Twelve years ago, it turns out, that this patient went to the hospital with her husband when she first started to get sick. They were prescribed risperidone, but she never took it. And you can see here the package is unopened. In this case, it was quite clear she was untreated. And then we utilized the adapted Matrix Consensus Cognition Battery, the MCCB, and then we went with -- Michael's done a marvelous job of adapting for this group and setting and do it under the best circumstances that we can, we administer it.
And then lastly, and critically, then the patient is referred to the treatment program, the national treatment program. So here you can see the psychiatrist talking to the patient, and the husband, that in the nearby village, there's going to be free medication provided in the next several months. She's already signed up if she wants to go.
I'm going to jump right into the results. This first analysis had 197 people with untreated psychosis in it. It was published in JAMA Psychiatry. Let me just walk you through the characteristics of the untreated group. Half are male, 57 percent Han versus the Huis, a sizable minority group within this province. They speak Chinese, Mandarin, so there's no problems in assessing them.
Ninety-two percent rural. Importantly, 52 years age on average. So again, in mid- to sort of later-ish life. Relevant to this workshop. Important also to the overall story, 3.8 years of education, so very poorly educated; 28.6 years of age of onset. This is actually typical in China and lower- and middle-income contexts. And lastly, you can see PANSS scores in different domains. Moderately sick, but not acutely so.
What is really surprising and extraordinary about this dataset, we believe, is that we thought the patients would be anywhere from 5 to 8-to-10 years untreated. It turns out that Michael's team is working in such remote areas of China that on average, as you can see here, they're over 20, on average, untreated, with up to 50 to 60 years of untreated illness. So this gives us what we believe to be an unprecedented opportunity to look at the natural course of untreated schizophrenia.
Let me walk you through an analysis in detail, and then I'll jump to the main findings. The MCCP has a number of subtasks. This one is the digit symbol coding. What the participant does is they match a nonsensical symbol to a number, it's a timed task, you have to do it as quickly as possible. That's the outcome of this regression model. You can see here that we have duration of untreated psychosis, so this is one year of untreated psychosis. And important to the story, we control for age of onset. Age, I'll get into this a little bit, there are two pieces: age of onset and then DUP, which we want to know the effective DUP, to control for age of onset.
We also, to do that we need to control for education, in this sample in particular, because even if they have a little bit of education, they do a lot better on the cognitive tasks. And then we control for gender, because there are gender effects in cognition. In this case, you can see that 1 year DUP is associated with worse performance in digit symbol.
But these are the main findings of the study. And you can see here in the adapted matrix that we use, there are six categories of domains of cognition, and then one or more test per cognitive domain. We also add to this battery, the mini mental status exam, because a lot of these folks are pretty poor performance on the cognitive battery. Michael also added this other variable. I thought it was a great addition. Basically, did they give even one valid response, one or more of these tests. And then you sum it, and there are nine of these tests, from zero to nine.
What you can see here, this is the takeaway, that for every year of continuing untreatment, that there's actually worse performance on number of valid test responses given, in processing speed, in reasoning and problem solving, and in visual learning. Per Bill Stone, we can understand that these domains -- we could interpret it as difficulties in executive function.
Again, coming back to the analytic strategy, we understand age can be partitioned into two pieces, age of onset and then DUP. We wanted to look at the effects to DUP, we found some associations intriguing. We do understand, though, of course, DUP is a bundled concept. It consists of normal aging, accelerated aging, psychotic illness itself, other processes. So we do understand we have to take a look at the other matched groups in order to really ascertain is this -- to contextualize this finding properly.
However, if it does hold -- we have a lot more work to do -- but if this does hold, it contributes to the emerging literature that the other two speakers have spoken about so nicely, around the possibility of continued cognitive decreases over time. Instead of stability, in terms of mid-adulthood and on, there could be continued decreases, as evidenced by our data. This was again, we published this in JAMA Psychiatry in 2020. It could be consistent with deficits in executive function.
Another note, we are engaged currently with another group called INTREPID. The PI is in London, and what they're doing is they are ascertaining people with incident psychosis, schizophrenia and psychosis, in India, Nigeria, and Trinidad. They have sort of a shorter DUP than our group, but we're really excited to partner with them and then to potentially even double or even triple the size of our cohort and be able to expand these questions further.
To sum up, there are a large proportion of untreated patients who remain untreated over time, and we see these cognitive decreases, again, initial untreatment finding. There are, just so you know, about 50 percent or even more of these folks do not accept treatment, even though it's offered and was free. What this provides is a scientific opportunity to follow them longitudinally. So we have a continuation R01 in order to do exactly this, and of course, as we assess them, we still try to offer them treatment. It's really up to them to take it or not, but we of course continue to offer it.
We're also trying to examine whether or not there is a precipitous cognitive decline, i.e., abnormal forgetting. That's indicative of dementia-type processes, so we can try to look at this again in the untreated course of schizophrenia. And then, in conclusion, we believe this really draws attention to this understudied group of rural, older, low-educated groups with schizophrenia, who we believe are neglected but important in low- and middle-income contexts and could constitute up to 30 to 40 percent of people with psychosis worldwide, and we believe that this is really -- we're leveraging this local opportunity within China to understand psychosis within the setting, but we believe that it could advance knowledge globally. If it does and these cognitive decreases do in fact reflect different pathophysiological processes than what we understood previously.
I want to thank you so much for your time and attention, and I want to turn it over to Dr. Elena Ivleva. Thank you.
ELENA IVLEVA: Thank you very much, Larry. Well, hello, everyone. First of all, I would like to thank the organizers for putting together such interesting and I think very important for the field workshop. I am delighted to be here and share some of my work on biomarkers of psychosis, specifically those which capture alterations in the anterior limbic system, along the schizophrenia course.
This is my presentation outline. I will start with a very brief overview of anterior limbic system and its contributions to cognitive function. Then I will show some of the examples of biomarkers, which focus on hippocampus, primarily, for the purposes of this talk, which occur at different stages of schizophrenia course.
And finally, I will take the opportunity to introduce our recently funded study, funded by NIMH, which looks at contributions of aging and illness course to lifespan neurobiology of schizophrenia.
To get us started, let's look into definition of limbic system. In general, the current thinking in the field is that the unitary concept of kind of a single limbic system is quite outmoded. Contemporary cognitive neuroscience studies refer to at least two systems within this limbic system, which includes anterior limbic system, comprised of anterior hippocampus, basal lateral amygdala, anterior cingulate cortex, and ventromedial prefrontal cortex -- primarily Brodmann areas 10 and 32.
In this component of limbic system is thought to contribute to specific processes within the global episodic memory construct, such as relational, also called associative, memory. This part of the limbic system also supports emotional processing and reward processing and their contributions to memory.
In contrast, middle/posterior limbic system includes posterior hippocampus, posterior and middle cingulate cortex, as well as fornix, myelinated body, anterior thalamus circuit. In this part of the limbic system makes contributions to different aspects of episodic memory, such as spatial processing, learning, and memory.
To delve a bit more in to the specific functions of the anterior limbic system, within the system, hippocampus and medial prefrontal cortex and their interactions are thought to play a critically important role in relational memory, and relational memory refers to binding of information into cohesive memory representations and their subsequent flexible application in familiar and novel environments. Relational memory sometimes is referred to as a core component of episodic memory, which in a sense makes episodic memory episodic.
Within this circuit, hippocampus in particular is known to support item-item and item-context binding, novelty detection processes, as well as pattern separation and pattern completion processes.
In medial prefrontal cortex supports other aspects of relational memory processes, such as source memory as well as self-referential cognition, including self-agency and broader self-referential cognition and sequence memory.
Importantly, in addition to this rather unique function of hippocampus and medial PFC, both of them contribute to broader memory systems, and as such, they're both part of the core recollection network, along with parahippocampal gyrus, posterior cingulate, retrosplenial cortex, angular gyrus, and middle temporal gyrus. So this network consistently activates during successful episodic retrieval across a wide variety of tasks, including associative retrieval, source memory, and remember/know tasks. And therefore, it is considered to be content-independent episodic memory recollection network. And this is a very elegant work from one of our collaborators, Michael Rugg.
Hippocampus in medial prefrontal cortex, as well as broader prefrontal cortex, is considered to be the key circuit in normal aging as well as in disorders of aging and memory, and this same circuit is also implicated in key aspects of schizophrenia pathophysiology.
So now I'm going to turn to the second part of my talk and show you a few examples of our work, focusing on hippocampal biomarkers across different stages of schizophrenia course. Starting with the hippocampal structure, reductions in hippocampal volume is one of the most reproducible structural findings in schizophrenia. It is a subtle reduction, so based on meta-analysis, it averages about 3 to 4 percent.
In one of our prior studies, we looked at whole hippocampal volume in a large sample of chronic psychosis programs, including schizophrenia, schizoaffective disorder, and psychotic bipolar disorder, as well as their first-degree relatives and healthy controls. And we defined hippocampus in the study through manual tracings as well as FreeSurfer parcellation. I must say, it was a heroic task, tracing near 1,200 hippocampi on these T1 images bilaterally. Took us about a year or so, but by the end of this, we found significant reductions in hippocampal volume in schizophrenia and schizoaffective disorder groups, based on both manual tracings and FreeSurfer, but not in psychotic bipolar disorder or relatives.
We also found direct correlations between hippocampal volume and total and verbal memory BACS -- brief assessment of cognition in schizophrenia -- scores, and inverse correlations with PANSS total and psychosis scores.
More recently, McHugo and colleagues from Stephan Heckers' group contrasted hippocampal subfield volumes, defined by FreeSurfer, in a sample of chronic versus early nonaffective psychosis and matched healthy controls. And they found reductions in both anterior and posterior CA regions of the hippocampus, not in dentate gyrus or subiculum, in the chronic psychosis group compared to controls, whereas in the early psychosis group, CA reductions were limited to anterior hippocampus only.
This data altogether demonstrates, first of all, that early stages of psychosis versus midcourse and perhaps advanced stages, are associated with different hippocampal structural biomarkers, and it appears so that hippocampal pathology in schizophrenia begins in its anterior compartments, and then spreads further onto more posterior regions as illness progresses.
In line with these differential findings in anterior versus posterior hippocampal structure, we examined resting state hippocampal connectivity, again, in this large sample, relatively large sample, of chronic psychosis individuals and healthy controls. And we used whole hippocampus as well as anterior and posterior hippocampi as seeds for this functional connectivity analysis.
We found a reduced functional connectivity between the whole hippocampus and several cortical and subcortical regions, including those within the anterior limbic structures, such as medial prefrontal cortex, anterior cingulate, and thalamus. And as you can see, these connectivity changes appear to be primarily driven by the anterior hippocampal connectivity, as the map of connectivity reductions was considerably broader for anterior versus posterior hippocampus.
What was also interesting here is that we did not find between-group differences in hippocampal connectivity between the three target diagnoses. So this may suggest that reduced hippocampal connectivity may be a biomarker of a broader psychosis dimension, rather than a specific DSM defined diagnosis.
In a recent study, Blessing and colleagues from Donald Goff's group looked at resting state hippocampal connectivity in a sample of first episode antipsychotic-naive individuals and healthy controls. And they used this very elegant data-driven analysis to derive hippocampal connectivity along its anterior-posterior axis. And what they found was, again, reduced connectivity between the left anteromedial regions of the hippocampus and several cortical regions, including insular cortex, anterior, medial, and posterior cingulate, as well as precentral and postcentral gyri. Furthermore, a subset of these individuals were rescanned after eight weeks of treatment with second-generation antipsychotics, and you can see that after eight weeks of treatment they showed near-normalization of this initial reduced hippocampal connectivity.
Yet another limbic biomarker we have been interested in is intrinsic hippocampal activity, which is regional activity at rest without specific cognitive task demands. To estimate intrinsic activity, we use ultrahigh resolution steady state CBV, cerebral blood volume, perfusion technique, called vascular space occupancy. The beauty of this technique is that it can be done with a very high resolution, and you can appreciate the voxel size here on CBV images as well as on structural images, and because of that we were able to look at not only the whole hippocampus but also at specific hippocampal subfields, delineated here.
And initially, when we compared our schizophrenia group as a whole and healthy controls, we found no between-group differences in CBV, any of the hippocampal regions. However, when we stratified our schizophrenia sample by illness duration into those who were early in the illness, less than 7 years, midcourse, 7 through 15 years, and advanced illness, more than 15 years, the pattern of hippocampal activity looked very, very different, with this consistently elevated CBV in the early psychosis group across all subfields, and reduced CBV in late stages of illness, and intermediate CBV in midcourse subgroup.
We found significant effect of group in the left CA 2/3, and a strong trend in the right CA2 and CA3. And these are the same data specifically from the left hippocampal CA3, output as (indiscernible), which gives us some sense of a trajectory of hippocampal intrinsic activity along illness duration, and it appears that hippocampal activity starts very high in early in the illness, then it drops nonlinearly and rather abruptly, within the first approximately 5 to 7 years of illness. Then it tends to plateau for quite some time, 10, 15 years, and then there is the second drop late in the illness, 25 or more years from illness onset.
And this brings us to the last part of my talk where I would like to introduce our current study, which just began this past summer. This study specifically focuses on in-depth characterization of anterior limbic system biomarkers, primarily focused on the hippocampus and medial PFC, as well as broader brain-based biomarkers, along schizophrenia course, as well as contributions of aging. The central question in this study is which aspects of schizophrenia lifespan neurobiology can be explained by the illness itself, which by abnormal aging, and which by interactions between the two? To support our predictions, we developed a working model of schizophrenia trajectories across the lifespan.
Within this model, we basically posit that there exists two dynamic but separate, even though complexly intertwined and interacting dimensions, in schizophrenia course. One is disease dimension, and second is aging dimension.
And within the disease course dimension, we predict that early in the illness we will find patterns of hyperactivity in the anterior limbic system, as well as perhaps in broader brain systems, while in advanced stages of illness, this hyperactivity will convert into hypoactivity of this specific circuit.
Perhaps the most interesting question here is this what will happen in midcourse, and when and how this initial hyperactivity will transform into hypoactivity? Within the aging dimension, we will be looking at accelerated cognitive and brain aging, based on a broad battery of cognitive and neuroimaging biomarkers.
Normal aging will serve as kind of a ground comparison dimension. Those small lines here refer to individual within-subject changes, because the study is designed as a longitudinal study with two-year follow-up, and I think one of the most important questions here will be to look at heterogeneity within this dimension and perhaps derive data-driven subgroups. Perhaps, for example, those with or without limbic hyperactivity, or with or without accelerated aging patterns.
Onto more practical aspects of the study, study design. Our planned analytic sample is 168 individuals, including 84 with schizophrenia and schizoaffective disorder and 84 healthy controls. We are sampling from a very broad age and illness duration range. And the study, as I already mentioned, incorporates this novel mixed lifespan study design, one of the novel designs currently used in the aging field. So the interesting feature of this design is that it allows for simultaneous modeling of continuous aging and illness duration effects, which will come in from cross-sectional data, as well as within-subject longitudinal changes, two-year changes in our case. This simultaneous modeling will allow for deriving most specific and precise trajectories for each of the biomarkers of interest.
In terms of the methodological approaches, it is a truly multimodal study. We have four different modalities of brain imaging, including structural MRI, perfusion MRI, fMRI, with a relational memory task, and both perfusion and fMRI tasks include ultrahigh resolutions from hippocampus and medial prefrontal cortex, and we'll also use magnetic resonance imaging to estimate glutamate, GABA, and other metabolites. All of this will be linked to cognitive function and clinical manifestations of the illness. We have fairly broad battery of cognitive assessments enriched for episodic and relational memory, as well as clinical assessments emphasizing illness course.
With this I would like to thank my collaborators and lab members from UT Southwestern Medical Center psychosis research program, as well as Advanced Imaging Research Center, and my collaborators from the University of Texas at Dallas, Center for Vital Longevity, who provide their expertise in aging. NIMH was supporting our study, and other agencies who supported my works through the years.
Thank you very much for your attention. With this I will turn it back to Dr. Michael Green, the moderator for this session.
MICHAEL GREEN: Thank you so much. That was really four outstanding presentations. I'm delighted to see them all together, forming an integrated package like this. At this point, I want to make sure all the speakers have their cameras on, and they do.
The first part, as before, is a Q&A. I'll make sure that each of you get at least one question, and then we'll move into the panel discussion. So let's start with Roman. You did such a great job of convincing me that there's an association between dementia and schizophrenia. So my question for you is why the heck did this get overlooked for so long? It seems like it should be obvious, but it wasn't. And how do you account for that?
ROMAN KOTOV: I think the real difficulty is just methodological difficulties in studying this question, because the onset of schizophrenia is, let's say, 25 on average. Onset of dementia, even in schizophrenia, would be 70, 75, on average. You, really to connect the two, have to do very long-term studies.
And you have not had a single study that tested cognition of more than a decade until about 10 years ago, when there was a follow-up of Marty Harrell's(ph.) cohort. But that one was 20 years and found no decline. But it ended at the average age of 44. So it's not just important to have a long study. It's also important to have a study that gets into later age, and so if you think back to the key slide I was highlighting, the loss IQ there was 1 point in two to three years, so typical follow-up of three or five years will not be powered, even if it is a couple hundred people, to detect it. So there are basically these limitations that only a good large and long longitudinal study can address.
MICHAEL GREEN: Thank you very much.
Eva, one of the things that I heard this morning from Dr. Payne was her comment, it was a brief one, but it was striking, about reality testing, in which she refers to the reality testing of seeing a facial emotion -- in this case a negative one like anger -- and then sort of asking the person what they meant. In this case, it was compensatory. It wasn't sort of the perception changed, it was the process of evaluating it changed, a kind of detective work.
How do you account for these affect-specific or valence-specific findings over time? Because there is a positivity effect over time. We tend to see, or disregard negative things. Is this is a sort of area of attack or relative strength for individuals with psychosis? What are your thoughts?
EVA VELTHORST: Let's say in general population that you can see what's going on there, because why it's mostly the fearful faces, for example, that there seems to be something going on there, instead of in the more neutral faces. Also, in the general population, there are several processes that have been suggested. I don't know if this is exactly what you meant by your question. But there's some general idea that for negative emotions, specifically to recognize those that might be more related to -- when you're older -- or sorry, younger people, they need to recognize, they need to be better to recognize certain negative emotions, because of their more openness to experience, or they seek fearful situations. And there's this idea that the older people get, this is a skill that they no longer need that much. There's also an idea, for example, that -- these are all theories out there, it's pure speculation, I don't know if it's ever been tested -- but it might also be that all these tests that have been done that look at these faces, that there might be a mismatch in age. So the faces are generally somewhat younger.
We tend to get along with people our own age more than people that are older or younger, so it might be there's some mismatch, and that may account for some differences in findings. There are several theories that are going on and that might account for that, that I don't think have ever really been explored in detail. It might also be that if you're young, you really have to rely on these cues more than when you're older, and when you know situations are safe or not safe. So I don't know if this is somewhat answering your question.
MICHAEL GREEN: It does. Thank you very much. Larry, we've got a couple of questions in, so we'll take the first one. This is from Gary Minogue(ph.). How do you understand that both DUP and antipsychotic medication usage being associated with poor cognition?
LAWRENCE YANG: Good question. I think, as I presented, I think the strengths of our study are obviously we're able to isolate the effects of the illness, because, again, just to be clear, these folks are completely treatment naive. They’ve never had any antipsychotics, as best as we can tell. So whatever decreases we see are independent of antipsychotic medication effects.
I would defer to my colleagues, Roman and Eva and Elena, who are more knowledgeable I think about the antipsychotic effects and the impacts on cognition. I think Roman did a great job of reviewing it.
ROMAN KOTOV: Just from a strategical perspective, one thing to, maybe one way to think about this is it's not duration untreated psychosis may be what's important, but duration of illness. And antipsychotics may help, may hurt. That's a separate question. But it is possible that a person on their own cognitive trajectory which may be normative for their age or may be accelerated, but then once illness begins, the trajectory changes. Or it even preceding overt illness, the trajectory changes.
In our data, this is Suffolk County project, age and illness duration correlate .5. It really depends on design of the study. You can end up with these two variables being effectively redundant or different. But it is possible to separate the effects, I think that's very important to study.
MICHAEL GREEN: Very good, thanks. We can come back to that in the panel, if we want.
Elena, I want to make sure you get a question as well. I really found the intrinsic hippocampal activity fascinating over age, over time, over the lifespan. At the same time, I was struggling to understand why it was hyperactive early, hypoactive later, when the impairments associated with that region do not show that same kind of pattern. You tend to see patterns across ages.
So do you have any ideas as to why the hypoactive-hyperactive is not reflected in performance-based measures, for example?
ELENA IVLEVA: This is a very good question. First I want to say, I hope it was clear from my talk, it's that we actually see very consistently in our work, as well as in others' work in the field, that hippocampal or broader -- hippocampal medial PFC -- biomarkers in early illness versus advanced illness, look quite different. So as I showed, anterior hippocampus comes up very early, and that is true in hippocampal structure. It is true in hippocampal perfusion. There have been also attempts to look into that through functional MRI tasks using high-resolution functional MRI. And then in later illness, it seems to be more of a broadly spread hippocampal and limbic pathology.
So first, I want to reemphasize that there seems to be a growing evidence that the limbic pathology does not stay static, and it actually changes quite a bit. Now, the question why it changes and how, and when are those inflection points in the life course trajectory, for each of these biomarkers, that is a $1 million question. And that's what we really don't understand. What's clear is that there seems to be this change, the kind of working theory in the field that it's been linked to regional excitation-inhibition disbalance, and specifically increased glutaminergic drive early in the illness, or perhaps reduced GABAergic inhibitory effects early in the illness, and reductions in both excitatory and inhibitory drive late in the illness.
So that's kind of the working theory, and there have been attempts to look into this in animal models as well in human MRS studies. However, this is something that definitely requires much more work, and it requires work in terms of mechanism, it requires work in terms of again developing trajectories. Because it's really not clear how this change occurs, if it occurs.
I think another very important part to it is that how is it that hyperactivity early in the illness but hypoactivity late in the illness contribute to seemingly kind of the same end-stage result? Meaning we know that cognition is impaired in early illness, and it seems like it's even more impaired in late illness, right?
Isn't it true that these hippocampal activity measures do show associations with both cognitive impairments, specific episodic memory impairment, as well psychosis severity. But what's really fascinating is that the direction of this change seems to change over illness course. Early in the illness, the more hyperactive the hippocampus is, the more severe impaired episodic memory, and the higher the severity of psychosis.
Where in late illness, it's the reduced hippocampal activity, impaired cognition, and more severe psychosis. So this is really intriguing, and it makes me think about parallels from aging, where, for example, in individuals with genetic propensity for Alzheimer's disease, as well as in early stages MCI, there is an association of hippocampal hyperactivity, which leads to cognitive impairment. But then as this progress progresses, in late MCI and Alzheimer's disease, it's opposite, where hippocampal activity goes down and then cognition goes down even further, and they're truly interrelated. So I think it's a very interesting question.
MICHAEL GREEN: Thank you. Very good, complete answer. And with this we transition to the panel discussion.
We have the cochairs, Ellen and Sophia, joining. I also see that there's a number of questions that the speakers have for each other, and so why don't we start with any questions from the two cochairs, and then give our speakers a change to go at it.
Agenda Item: Panel Discussion with Chairs
ELLEN LEE: Thank you so much for the wonderful presentations. I just wanted to add on, because I know there are a lot of interesting questions about the sampling for these studies, and there are a lot of questions about managing cultural differences, and I wanted to give some of the panelists, and in particular Larry, a chance to respond to some of that.
LAWRENCE YANG: I guess I will start off. In brief, the sampling from our study is through this treatment program, just to describe a little bit more. These folks are actually identified by what are called, quote-unquote, village doctors. They're not really doctors, the way we think about it. They're more like community health workers. But they're folks who live in the communities, in these villages, and then they recognize, they're given a description of people with long-term psychosis, and then through them, then they're identified for the treatment program. That's how we ascertain them.
There's definitely a bias towards finding folks who have long-term, disruptive, some kind of impairment, because otherwise, how would we find them? The village doctor would not be aware of who this person is or where they live. So there's certainly -- our study definitely is skewed towards people with schizophrenia that's chronic in nature. We do not have people who, for instance, have psychosis or schizophrenia and then remit.
The second question is a great one, around context. What I would say is that Michael has done just this incredible job of having to modify the matrix battery for use with really lower functioning and very low educated people because the matrix was designed for people fifth grade and up, and that makes a lot of sense, especially given the cognitive tasks. So there were, without getting into too many details, there was Michael sort of added a lot of pieces, didn't change the test itself at all, but sometimes he would modify or adapt training to the test so that the patients would -- I'll give you a very clear example. Some of tests use a computer mouse, and some of the patients had never seen a computer before. So obviously the person needed some orientation around how do you move the mouse, et cetera. So there was additional training according to the content.
MICHAEL GREEN: Thank you. Sophia, did you want to also ask a question before we go to the speakers?
SOPHIA FRANGOU: Yes, I would love to. I have a couple I'll start with, if I'm allowed. So there's a line of analysis that have been done looking at cognitive subtypes within schizophrenia, and I know that a very large European study, the PRONIA study which was essentially focusing on prodromal presentations, found two groups, what they called a spared and an impaired group, for obvious reasons.
And then we haven't published this data yet, but we just finished this analysis. We looked at PSYSCAN, another large international consortium of first-episode patients, and we find again the same thing, an impaired and spared one. And then we also looked at a U.S. sample, about 200 people in it, first-episode presentations, and again a spared and an impaired one.
And all the presentations that we had so far suggest a much more uniform cognitive profile. So I'm trying to reconcile the two things in my mind, and I would value input from people have been doing that for a while.
MICHAEL GREEN: A number of people could answer. Eva, do you want to make a first pass at this response?
EVA VELTHORST: I will try. So you're talking about cognitive profiles that are either generally impaired or generally non-impaired. I think this is -- how to make sense of that? Well, I guess this brings us back to the idea of Dwight Dickinson that says one sort of underlying general cognitive impairment which sort of explains or drives the cognitive impairment overall, and that that might be more or less present in individuals, but there's also -- even for people with the highest level of cognitive functioning, it's now been thought that they are also not spared. Even though it might appear, when they sort of like conduct tests and they might function better than the rest, but we don't know what their premorbid functioning is.
So there's now also a line of thought that says, well, are they really spared? We're just looking at test score, because we really need to consider their premorbid functioning, even though they do well on these tests, they did way better before. So yes, you find spared and more impaired test, and I think that really has also clinical consequences, but I'm not sure whether they were always spared or whether even the spared group may have experienced a level of decline prior to illness onset. So I don't know how that answers your question, but obviously there's heterogeneity between individuals because they come from different starting points, and obviously not everyone impairs to the same extent. It also depends on multiple other factors, like illness severity overall and negative symptoms and all that. That's my stab.
MICHAEL GREEN: Excellent, Eva. Anyone else want to comment? This is an area in which there's been a fair amount of discussion. Roman, did you want to?
ROMAN KOTOV: Let me quickly just on the statistics of it say that one thing to a child with class analysis, which I'm not disputing as a general tool, but just a pitfall of them is that they are designed to produce classes. Very rarely would they (indiscernible) in my experience, the conclusion would be, no, just one class. Because there's some heterogeneity, the model is trying to deal with it, it's producing classes to deal with it.
It's important to have as a comparator whether this is cross-sectional analysis, longitudinal analysis, a dimensional account of the same variants, and see which ones fit the data better. In my experience, it has been dimensional. That is, if you look at distribution of cognitive scores, you will see gradual histogram of something like this result, these two groups that latent class analysis identify actually been present to the naked eye, but really the formal tests of course don't determine whether there is something underneath the histogram distribution. It really is just latent trait.
But I want also to circle back to the question of samples and to say that, you know, just it is scary to me to look at the data for people who are old and have schizophrenia and draw inferences, which of course we must, but draw inferences about schizophrenia, because who is in those samples?
In our data, for example, we have people who we know they are representative of at least those treated for schizophrenia at least admitted to the hospital and now we know how much care they are getting. Eighty-six percent of them and then you cross-section and then you want (indiscernible) have not been hospitalized. So if the detection or catchment area is defined as hospitalization with some 55, and in a given year interval, it would pick up only 14 percent of those who have lifetime diagnosis.
Even if you take people who you have in mental health that is outpatient services, still there is at least a quarter who are not engaging them, some of them better, many of them just avoiding services, even living in the woods quite literally, and we would be missing them without sort of a kind of frame that would ensure that we can find everyone.
MICHAEL GREEN: That is good. In fact, Roman mentioned the point that I was going to comment on, which is the methods themselves lead to this, but the other thing about this session is it does a good job of pointing out that cognition ain't the thing if you start looking at social and nonsocial cognition across the lifespan. One of the most depressing things is to look at how quickly working memory and speed of processing drops off early in adulthood but what keeps many of us going is the fact that social cognition does not. So then when you sort of look across the lifespan and Eva's talk is a good example, you might find different patterns. Of course we're interested in interactions by diagnosis, but they're just simply different enough that they have different time courses.
To the chairs, could I open it up to the speakers for questions at this point? I have several. Roman, you started off with one for Larry.
ROMAN KOTOV: Sure. So, Larry, my question for the audience is it would be another critical piece of information to see how difference between age made psychiatric individuals changes over time either by illness duration or by age.
LAWRENCE YANG: Absolutely. So, thank you for your comment. As I interpret it -- exactly. The data that I showed you today was just in the untreated group, right? So in this group we can understand their illness duration is their duration of untreated psychosis, because they've never been treated. So those two things are synonymous.
However, in the matched treated groups that we do have a sample that's matched on age band, education, by gender, or ethnicity, that we found folks who are as comparable as possible, you know, a full task, but as comparable as possible. These folks obviously would have a DUP, but then they would also be treated, right? So then there would be a duration of treated illness that might be interspersed with nontreatment, et cetera.
But we would be able to look at compare in the treated group this sort of DUP plus duration of illness, treated illness, and compare it to the untreated group, which is all untreated illness presumably. It's a great suggestion. It's the next -- I think that's probably the next thing on our list. So stay tuned. Look forward to having further dialogues about this.
MICHAEL GREEN: Thank you. Eva, you had a question for Roman.
EVA VELTHORST: If I may. Obviously, of course, I am stunned by these results. I think they are very interesting or spectacular, because as you know, we looked at the 20 years trajectory and there was some level of decline, not that much. So my idea always was, okay, this is it. It's just very gradual. I thought maybe they reach a certain level at some point where older individuals are maybe misdiagnosed with dementia, because their cognitive reserve, they reach this below point.
First, I was wondering about your thoughts on that, but another thing I was wondering about whether you found -- whether you were able to look at whether the cognition patterns were a bit different at older ages, whether you found different types of cognitive impairments, just to see whether it might be completely or not completely -- but it might not be the same gradual process as pure cognitive decline with another sort of like qualitatively different dementia process that is going on.
ROMAN KOTOV: I think you are asking two questions. Let me start with one I can answer quicker, that different patterns of change, we couldn't really do this here, because we really wanted to have premorbid course as part of this trajectory, and premorbidly it was mostly full-scale IQ. Sometimes the score separately from verbal versus performance and sample size started getting quite small there.
But we did look at performance IQ by itself, just from year 2 of the study to 25, and the trajectory looked exactly like the one I showed of basically full-scale IQ. However, if you look at vocabulary, vocabulary doesn't show this up until year 20, but after year 20, it starts declining. That's significant, because vocabulary is quite protected from cognitive changes. So that's my short answer. The longer piece is that can we model this nonlinear change? In fact, we tried that in the data, again, in the full trajectory that I showed. It was really linear for the sample overall. We didn't need like quadratic term there.
But individuals could be following a rather different course. So the group is composed of hundreds of individuals, and the average decline could be because one person goes flat and then declines relative to everyone. The other continues going flat, and then declines rapidly later. But in order to search for this turning point on individual level, you need to do them, have to have more observations for people than we had. Maybe like six or more. And perhaps (indiscernible) you'll eventually get to that number of follow-ups.
MICHAEL GREEN: That's great, thanks. We have a few minutes left. This question and answer, relates to something that has come up several times, which if I were to be provocative, I would say is the concept of accelerated aging the correct term? Is that the correct concept anymore?
Because there seems -- and I think Eva used the term second level of decline. Most of the data we saw at the group level was nonlinear. So is that sort of a useful entree concept but now we have to say, well, something happens between these decades, but then something completely different happens at these decades? And I think all of you had work that sort of touched on that.
Any thoughts about the usefulness of the concept?
EVA VELTHORST: Well, not really, just only that I agree with you that, well, from the data up to the age of 50 at least in cognition and as we see in social cognition up to what we know thus far, it does not seem to be accelerated. It seems to be fairly static up to a certain level after which there starts to be something completely different going on. So it might -- well, maybe it may still be accelerated, but it only starts very late. It's just a different process than the gradual decline that happens beforehand. I don't know.
ELENA IVLEVA: I will also add that in our hands, what's really interesting is that different biomarkers behave very differently in terms of disease course effects in accelerated aging patterns for example. So all this is fairly preliminary, but functional measures like for example regional perfusion or in fact MRS data appear to be more sensitive to illness course effects and not so much to aging effects. At least when you try to kind of cover out the other.
While structure, for example, looks fairly similar between aging effects and illness course effects. So I guess I am okay with the term, but understanding all the complexity behind it, including individual heterogeneity as well as this complexity around measures and sensitivity of different measures to these processes.
MICHAEL GREEN: Thank you so much. That brings us to the end of the session. I want to thank the speakers. I want to thank the cochairs for a session that was not only informative; it was downright fun. So this is really a terrific series of presentations. I'm going to ask the speakers and the cochairs to turn off cameras and to mute, and we have a 15-minute break starting now. Thanks, everyone.
JUDY FORD: Good afternoon and good evening to those of you far east from me. My name is Judy Ford. I will be the session moderator for this final session of today's meeting entitled Regulatory and Metabolic Processes.
It is my pleasure to introduce the four speakers. The first one will be Dr. Bengi Baran from the University of Iowa. The second will be Dr. Ellen Lee, from the University of California in San Diego, followed by Jill Glausier from the University of Pittsburgh, followed finally by Dost Ongur from McLean Hospital and Harvard University.
As in the previous sessions, each speaker will have 15 minutes for their presentation. At the end of those four talks, we will have a 10-minute Q&A and at the end of that, we will have a panel discussion where we will pull in the chairs, Ellen Lee, who will already be on, and Sophia Frangou.
So let's kick it off with your talk, Bengi.
BENGI BARAN: Thank you very much, Dr. Ford. I am delighted to be the first speaker of this highly interesting and multifaceted session. I'm also excited that the workshop brings together two populations that have been the focus of my research and inspires me to integrate them. I typically like to start my presentation with all the spoilers ahead of time. So today, I will briefly talk about the mechanism by which sleep enhances memory consolidation and then move on to age-related changes in sleep oscillations and their relationship with cognition.
Then I will move on to my postdoctoral work in Boston with Drs. Dara Manoach and Bob Stickgold, who have identified a very specific sleep oscillation sleep spindles as a putative endophenotype of schizophrenia. Finally, I will finish with some promising treatment strategies.
Before I go any further, let me start by quick fact that you already know. Sleep is not a unitary phenomenon. Rather, sleep physiology keeps evolving throughout the night. Different patterns of EEG activity differentiate sleep stages. These sleep stages are also associated with distinct brain states, such that neurotransmitter concentrations and direction of communication within brain structures changes across REM and non-REM, creating the ideal electrophysiological environment to consolidate memories.
And memory consolidation is the process by which recently acquired memories are transferred from the hippocampus to more stable representations in the cortex. This process takes place offline, and it's favored in sleep. The research on sleep's role in cognition burgeoned exactly 20 years ago with this cover feature on Science, and since then has been gaining a lot of attraction and interest.
So in a typical experiment at the behavioral level, what we observe is that memory performance across a wide variety of tasks is better if learning is followed by a period of sleep compared to an equal amount of continuous wakefulness.
In this rodent study pioneered by Matt Wilson at MIT, we were able to get a glimpse of what's happening in the brain during sleep. So with this very simple and cartoonlike demonstration, what I am trying to show is that if you record hippocampal activity in the animal during a successful maze run and identify the patterns of neural firing during learning, and continue to record -- continue with this recording during subsequent sleep -- you will notice that the same neural assemblies that are activated during learning are reactivated during sleep, and remarkably in the same order.
So even though the animal is sleeping, the hippocampus is replaying and practicing what it just learned during wakefulness.
In fact, we now know that memory replay during sleep relies on three cardinal non-REM oscillations that play the critical role in their transfer from the hippocampus to the cortex. These are thalamocortical spindles, hippocampal sharp wave ripples, and cortical particularly, prefrontal, slow waves. It's the exact precise temporal coordination of these three oscillations are nested within each other that brings about memory replay during sleep.
So this is all great news, right? Sleep is nature's magical pill that improves cognition. But unfortunately, sleep changes throughout development. With aging, we observe a reduction in total sleep time, which is accompanied by difficulty in falling asleep and also difficulty in maintaining sleep. There's also a gradual decline, gradual reduction, in slow wave sleep, which is critical for memory consolidation. So aging also changes sleep architecture. Whereas young adults show uninterrupted large chunks of each sleep stage, sleep in older adults is likely to be more fragmented.
So given these macrolevel changes, we first sought to identify how aging influences sleep-dependent memory consolidation. We found that older adults get a sleep benefit for some memory tasks, but not for others. We observed a decline in the sleep benefits of a procedural sequential memory task as early as middle adulthood. But not on a declarative associative learning test.
And even when we focus on this relatively preserved type of memory, declarative memory, I conducted a study in which participants learn a task outside the scanners, then they had a 90-minute nap opportunity, followed by a completely new recall task inside the MRI scanner, and what we found was that there was a significant negative relationship between frontal slow wave activity during the nap and hippocampal activation during recall in the subsequent recall task in young adults in blue.
So those who had a nap rich in slow wave activity needed to recruit the hippocampus less during recall. On the contrary, this was a positive relationship in older adults.
Then we examined task-based connectivity and found increased positive connectivity between the hippocampus and two clusters in the right and left prefrontal cortex during recall in older adults. We interpreted this as a failure in older adults to transfer memories from the hippocampus to the cortex during the sleep opportunity and the compensation mechanism that still highly relied on hippocampal activation and hippocampal frontal connectivity.
Further work related this to brain structure. Age-related cortical atrophy, particularly of the medial prefrontal cortex, correlates with slow wave activity. It also correlates with sleep-dependent memory consolidation, and I apologize for this busy set of figures. What I'm trying to show here is that sleep microstructure deficits in aging also extend to sleep spindles, and it's not just the reduction in sleep spindles or some oscillations but an abnormality in the precise timing of the two of them with each other that predicts deficits in memory consolidation.
So let me just very rapidly switch gears and talk about the schizophrenia story. This audience needs no convincing that cognitive impairment is a lingering symptom, even when positive symptoms are under control, and plays a significant role in reducing functioning in patients. The work I will briefly summarize for you here is focused on sleep-dependent aspects of these cognitive deficits in schizophrenia.
Patients with schizophrenia have a specific reduction in sleep spindles, and this importantly occurs in the context of fairly normal sleep architecture. This reduction in sleep spindle density correlates with sleep-dependent memory consolidation. What I'm plotting here is overnight change in performance on a procedural memory task. While performance was enhanced in controls in blue after a night of sleep, on average there was no improvement in the schizophrenia group, and although as a group average there is no improvement, when we examine the range of performance in patients, the spindle deficit significantly correlates with MSD performance.
This finding of reduced spindle activity is well replicated. It correlates not only with deficits in sleep-dependent memory consolidation but also with symptom severity, and importantly, it's been ruled out that it's not due to disease chronicity or a side effect of medication in patients. It is also observed in antipsychotic-naive early course patients, as well as first degree relatives.
And my primary interest has been the circuitry that produces spindles. Spindles are initiated by the thalamic reticular nucleus and are propagated to the cortex and synchronized through thalamocortical feedback loops. Speaking of thalamocortical loops, as revealed by resting state connectivity MRI, which in its simplest form takes the thalamus as the seed and compares its connectivity with the rest of the brain across patient and control groups. This approach reveals a pattern of abnormal hyperconnectivity and such that patients with schizophrenia have increased connectivity of the thalamus with the sensorimotor cortices. This finding of abnormal thalamocortical connectivity is also well replicated, and importantly it's observed in clinical high-risk and predicts conversion.
In a recent study, we asked a very straightforward but informative question. Are these deficits related to each other? We replicated previous findings of reduced spindle density, this time using a high density EEG array. We also replicated abnormally increased thalamocortical connectivity with abnormally-increased sensorimotor thalamocortical connectivity in a separate MRI scan, and then looked at how they related with each other.
And we found that the regions of the thalamocortical network that showed relation to its spindle density largely overlap with the deficit region, the hyperconnectivity region, such that those who show the most reduction in spindle density in turn are those who show the highest sensorimotor hyperconnectivity.
We argue that these deficits are related because of an underlying shared pathophysiology. I mentioned that spindles are generated by the thalamic reticular nucleus and propagated to thalamocortical connections. The TRN is entirely (indiscernible). It's made up of GABAergic neurons. So I can borrow one of my colleague's terms, such that a leaky thalamic reticular nucleus would fail to propagate spindles during the night and be responsible for deficits and sleep-dependent memory consolidation.
But can we increase sleep spindles in patients? The short answer is yes. We completed the clinical trial in which we administered a non-benzodiazepine hypnotic agent to patients with chronic schizophrenia as well as matched controls, and even with just a single dose we were very successful in increasing spindles. Red here means increasing spindle density compared to the placebo condition, and both groups showed this increase, but more so in patients such that the spindle density on the drug night was quite comparable to that of the controls.
But there's a catch. There's always a catch, right? We examined sleep-dependent memory consolidation across drug and placebo nights, and if you look at only the healthy controls here in blue, we didn't see much of a difference between those two conditions. But patients got worse. This was quite perplexing at first blush. Why would you see a worsening of memory consolidation if you're boosting the spindles, right?
So then what we did was we examined what the drug did to the precise temporal coordination of sleep spindles with slow oscillations, and what we found that although we successfully increased boosted sleep spindle oscillations, the drug interfered with their temporal coordination, the cortical slow oscillations, and that's why we did not observe the desired effect on memory.
So hopefully at this point, I'm hoping that I managed to bring both of these stories together. Both aging and psychotic disorders interfere with the process by which we consolidate and strengthen newly acquired memories. And this gives us a promising treatment target. But also it's a challenging one. We can increase these oscillations with drugs or with stimulation, but the challenge is to increase them while also maintaining their temporal relationship.
That's why I wanted to end on a promising note of newly emerging treatment strategies. Using a closed loop approach, if we were to use a closed loop approach, we can detect oscillations and target them. For instance, this is just the proof of concept study and it's only in healthy young adults. But the authors used transcranial alternate current stimulation in order to increase spindles. So they would detect spindles and then they would stimulate at the spindle range, and therefore boost them.
But this approach fails to address the issue that I just mentioned, the coordination, controlling for the coordination between oscillations, and at that point, I would like to point, highlight the work of my colleagues at Mass General, which is still in progress. Auditory stimulation with burst of pink noise also increases spindles, and because there's no electrical stimulation, you're able to monitor the changes the stimulation exerts on sleep oscillations in real time.
And using a closed loop approach, my colleagues are working on detecting slow oscillations and they started the auditory stimulation to produce a spindle exactly when it should appear to trigger memory replay. If proven successful, these precise interventions can be scaled up for instance by the use of ambulatory EEG monitoring technologies, wearable technologies, and can be used in both psychosis and aging.
Thank you very much for your attention. I would like to highlight NIMH funding that made this work possible as well as the valuable contributions of my mentors, my colleagues, and also my trainees, and at this point, I would like to pass the torch to Dr. Ellen Lee who will continue within the story of sleep, and I'm looking forward to her presentation on obstructive sleep apnea. Thank you.
ELLEN LEE: Thank you for staying with us. I'm going to talk today about some of our work in obstructive sleep apnea and schizophrenia.
These are my disclosures. Much of this work has been supported by grants from the National Institute of Mental Health, as well as from the Brain and Behavior Research Foundation and the American Psychiatric Association.
So as Dr. Baran so eloquently put before, sleep disturbances are really common and undertreated among people with schizophrenia. About 30 to 80 percent of adults with schizophrenia have trouble with sleep latency or falling asleep, overnight awakenings, as well as poor sleep quality.
Many of these sleep problems actually precede the onset of the psychotic symptoms, the development of subsequent cognitive impairment, and the initiation of antipsychotic treatments. Sleep problems are also associated with lower quality of life, worse cognitive functioning, and poor physical health.
Importantly for me, sleep disturbances are also underdiagnosed and undertreated in people with schizophrenia and are an important modifiable target for intervention. In some of our earlier work, we found that people with schizophrenia who have poor subjective sleep quality also have higher levels of inflammation.
Among this group of 144 people with schizophrenia, we found that high sensitivity C-reactive protein and interleukin-6 levels were linked to worse sleep quality as well as worse cognitive functioning, controlling for both sex and age. This sample actually did range in age from about 26 to 65 years old. Over here on the plot, we're seeing interleukin-6 levels with the poor sleepers shown here in red, and then the good sleepers shown here in yellow.
Extending upon this, we started to look at objective sleep measures using actigraphy in people with schizophrenia. So this is showing just a small sample of 43 individuals with about 427 nights between them. This is a plot showing total sleep time, in minutes, on average, over the five to seven nights of sleep between the nonpsychiatric comparison subjects shown here in blue and the schizophrenia group shown here in red.
We see that the total sleep time on average is very similar between the two groups. However, when we look at the night-to-night variability of total sleep time using root mean square successive differences, which really captures how sleep varies consecutively as opposed to using a standard deviation which will sort of cancel out the proximity of nights. So order matters. If you sleep badly on night one and then you sleep better on night two, the variations are really what's picked up here.
We see that the variability, the intraindividual variability of total sleep time is much higher in the schizophrenia group compared to the nonpsychiatric comparison subjects. We found that across, actually, most objective sleep parameters, including sleep efficiency, wake after sleep onset, and bed time and wake time.
So while I'm very interesting in how sleep disturbances influence cognition and functioning, one of the key confounders of my work here has been the presence of obstructive sleep apnea that is underdiagnosed and undertreated.
So obstructive sleep apnea, or OSA, is a very common condition worldwide. It affects about 936 million adults between the ages of 30 to 69. The incidence, the severity of OSA increases with age. The overnight collapse of our airway that then leads to this intermittent what we call apneic events or hypopneic events, where peoples' oxygen saturation dips to crucial levels actually has many, many cascading effects both in terms of oxidative stress, inflammation, parasympathetic activity changes, that influence health including cardiovascular disease and other outcomes.
Among people with schizophrenia, while the literature is fairly new in this area, we see that about 46 to 62 percent of certain populations with schizophrenia have obstructive sleep apnea. One study found that compared to the general population, people with schizophrenia were about twice as likely to have obstructive sleep apnea. Among people with schizophrenia, we found that OSA is associated with depression, negative symptoms, and antipsychotic medication uses. So higher doses, or more frequent, or longer durations.
Ultimately, one of the important things is that untreated OSA can actually impact is the effectiveness of insomnia interventions. So the risk for obstructive sleep apnea we did find to be increased among the people with schizophrenia in the cohort that I was studying. So this is showing results from 82 people with schizophrenia, shown here on the right, and then about 94 nonpsychiatric comparison subjects.
We used a subjective reporting measure that assesses sleep apnea risk. It's called the Berlin sleep apnea questionnaire. While not perfect, we did find that there were some interesting findings of risk between these two populations. About 46 percent of our schizophrenia group, shown here with this orange bar, had high risk for sleep apnea, compared to about 21 percent of our nonpsychiatric comparison subjects.
We found that among the people with schizophrenia, sleep apnea risk was associated with worse depression, worse sleep quality, and increased perceived stress levels. We also found that sleep apnea risk was associated with increased inflammation as well as with medication dosing. So the far left plot is showing high sensitivity C-reactive protein, or hs-CRP levels, between about 53 individuals with schizophrenia, subdividing the high risk group shown here in orange and the low risk group shown here in green.
We're seeing higher levels of hs-CRP in the far left plot. In the middle plot, we're looking at another cytokine called interferon gamma. We're again seeing that the schizophrenia group at high risk for OSA has higher levels of interferon gamma. And in terms of daily antipsychotic dose, our higher risk group has higher daily antipsychotic dosing based on the WHO definition of equivalence, compared to the low risk group.
So while this is all very interesting, we know that sleep apnea risk assessments are not always perfect in schizophrenia. There was a recent paper that showed that there are different subjective measures to assess sleep apnea risk and ultimately some are not as sensitive in people with schizophrenia.
So we turned to looking at objective assessments of sleep apnea. In this study, we looked at one-night home sleep testing. An example of that test is shown here. It's a clinical grade assessment. It involves a pulse oximeter worn on the finger to assess blood oxygen saturation, a nasal cannula to assess air flow, and a chest strap to assess for respiratory effort. The study is read by a sleep technician and can provide information about the presence and severity of obstructive sleep apnea.
In this pilot study of home sleep testing, our sample included 13 people with schizophrenia and 11 nonpsychiatric comparison subjects. The age range in this sample was 30 to 68 with a mean age of about 51.7. About 45 percent of our sample were women.
The two diagnostic groups actually had similar self-reported sleep quality measures. We excluded any individuals with a diagnosis of obstructive sleep apnea or who had been previously treated for sleep apnea.
In terms of feasibility, we had deployed 71 devices. About 56 individuals were able to complete the study. We derived useful data from about 24. The mean monitoring time was about 5 hours and 59 minutes. And while our sample is really small, I did want to show some interesting findings that were being suggested with the relationship in terms of age.
In this sample, on this plot I'm showing age across the x-axis and the apnea-hypopnea index which gives us a sense of a number of events of apneic or hypopneic events per hour, which is the measure for both diagnosing obstructive sleep apnea as well as assess the severity. The horizontal line across here is showing an AHI of about 5 which is the cutoff for the mildest of disease. Higher cutoffs of 15 and 30 are used for both moderate and more severe sleep apnea.
The red dots here are showing the individuals with schizophrenia while the blue dots are showing the comparison subjects.
In general, we are seeing that those in our sample who are in the older age range are shown to have higher severities of obstructive sleep apnea. In general, we're also seeing that the AHI is higher in the schizophrenia group versus the controls.
So we took a quick look at how severe these events were in the different groups. As we mentioned before, the apneic and hypopneic events were much more common in the people with schizophrenia. This plot on the left is showing the AHI. The people with schizophrenia in the red and the nonpsychiatric comparison subjects are shown here in blue. We're also seeing that the average overnight oxygen saturation is much lower in the schizophrenia group compared to the controls, and that the duration of these desaturations, so we defined it here as less than or equal to 90 percent, is much longer in the schizophrenia group relative to the controls.
Overall, when we assessed who met criteria for at least mild or more severe disease, we found that 46 percent of the small sample met criteria -- of the small sample of people with schizophrenia, and about 36 percent of the comparison subjects met criteria for obstructive sleep apnea.
Now, while we were limited due to the small sample size, we did want to explore how OSA was related to psychopathology. This very small group, we did find that people with schizophrenia who also had OSA had worse positive symptoms, worse negative symptoms, higher antipsychotic doses, and actually better self-reported sleep quality compared to the people with schizophrenia who did not have OSA based on objective measures. This is really intriguing to us because this is something we'd like to examine further both in terms of whether or not this holds up with a larger sample as well as to understand how sleep apnea treatments could influence symptomatology as well as functioning in these groups.
The interesting thing about the better self-reported sleep quality was that we realized that many of the individuals who were not found to have obstructive sleep apnea did have poor sleep quality. So there may be some comorbidity of insomnia in this group that is not being addressed. So really understanding how insomnia, obstructive sleep apnea, and then the combination of having both comorbid insomnia and obstructive sleep apnea may affect outcomes differently among people with schizophrenia.
So just to touch a little bit on some future directions that we think warrant further study is: how do antipsychotic medications and lifestyle factors, such as diet and exercise and smoking, contribute to obstructive sleep apnea risk? Many of these things are inherent to our population of aging people with schizophrenia, and yet it's important to think about how to improve their risk and ultimately improve functioning in this subset.
Also, do people with schizophrenia who also have OSA have worse psychopathology or cognitive functioning than people with schizophrenia who do not have OSA? Are there subgroup differences or are there ways that we can actually improve the deficits that we see because they are related to actually the sleep apnea as opposed to just being inherent to having schizophrenia itself?
That leads me to my last point. Does obstructive sleep apnea treatment among people with schizophrenia lead to clinically meaningful improvements in health and cognitive functioning in this group? That is an area that we believe warrants further study.
Just to close, I want to thank the many participants who donated their time and energy to be part of our study as well as the incredible group that I work with who gets a lot of this done, as well as my mentors and collaborators shown here on the slide. Thank you so much for your time and attention.
I'd like to hand it over to Dr. Glausier.
JILL GLAUSIER: Thank you, Dr. Lee. Okay, in a bit of departure from the rest of the presentations today, I am going to present a transcriptional analysis that utilized postmortem human brain tissue in order to begin to assess mitochondrial functions in the schizophrenia brain.
The dorsolateral prefrontal cortex is a locus of in vivo functional abnormalities in schizophrenia. DLPFC dysfunction is thought to underlie, at least in part, the pervasive and persistent cognitive deficits experienced by people with schizophrenia, and we heard a lot of great talks in the cognitive section today about that. Many of the functional, morphological, and molecular cortical alterations in schizophrenia have been associated with this idea of mitochondrial dysfunction.
Now, if saying mitochondria brought you immediately back to middle school thinking of it as the powerhouse of the cell, you're exactly right. So the primary and arguably most important function of mitochondria in our brain is to synthesize ATP. They do this primarily by a process known as oxidative phosphorylation.
This is shown schematically here as a little schematic of a mitochondria. Within the cristae membrane are the five complexes of the electron transport chain through which oxidative phosphorylation occurs.
In the brain, nearly all neuronal ATP is synthesized via OXPHOS and the majority of that ATP synthesized goes to supporting neuronal firing and synaptic neurotransmission, so basis of neural communication. However, mitochondria also participate in a number of other functions in neurons, and I've just listed two here that are particularly important.
For example, mitochondria are critical for calcium buffering, and this is particularly important for mitochondria that have been trafficked to the presynaptic bouton. Mitochondria are also important for apoptosis and apoptotic-like processes. For example, processes involved in spine pruning. Findings from both living subjects and in postmortem brain tissue show perturbations to one or more of these mitochondrial functions in schizophrenia.
However, there is still this really big outstanding question, which is what is the nature of mitochondrial alterations in the schizophrenia brain? And to begin to chip away and answer this question, we performed a transcriptomic analysis of DLPFC brain tissue. We did this in a curated list of genes whose purpose was to index this diversity of mitochondrial functions.
We had three main goals. First is to examine and quantify the severity of mitochondrial alterations. Next, we wanted to look to see if those alterations were enriched for any specific functional pathway. Finally, we wanted to look at the higher-order gene coexpression relationships.
In order to curate this list, we went to gene ontology and selected the pathway mitochondria. So I'll call this GOMito from here on out. There are a little over 1,000 genes in this pathway. So essentially any gene whose product is there is some evidence that it participates in mitochondrial function is included in this list. So we really do see a broad diversity of mitochondrial functions being represented.
These analyses were performed in the DLPFC on total grey matter. Here is an example of a section through a postmortem human brain in the coronal plane. The DLPFC is around here. It's actually a pretty large area of the brain. The grey matter was excised from the DLPFC and then sent for RNA sequencing that was completed as part of the CommonMind consortium.
This was completed on individuals who had schizophrenia when they were alive and then unaffected comparison subjects who had no evidence of any major neurological or psychiatric disorder during life. Of those 1,033 GOMito genes, RNA sequencing was able to identify 871 of those.
So I want to pause for a minute and go over the subject characteristics for the individuals that were included for analysis, and in particular, the mean age of these participants. For both groups, unaffected comparison and schizophrenia subjects, solidly in middle age, and so the results that we're getting today are from individuals who have had schizophrenia for a while and who are in, by and large, midlife or beyond. These three measures below give us some feel as to the quality of the tissue that we're looking at. All of the numbers, mean values, that we see here represent excellent preservation, and importantly they don't differ between the two groups.
Finally, we see representation of both groups, both white and black subjects, male and female subjects, and in total there were 82 unaffected comparison subjects and 57 schizophrenia subjects.
This volcano plot shows results on the x-axis of the schizophrenia disease log twofold difference, on the y-axis, the negative log10 p-value, and the dotted line shows a Q value of 0.05. Each one of these markers is an individual GOMito gene. So every marker that's above this dotted line was statistically differentially expressed, or a DEG, a differentially expressed gene, and 41 percent of those GOMito genes were identified as differentially expressed in individuals with schizophrenia.
Just a qualitative look at the volcano plot shows that we have some genes that were upregulated in expression and many more that were downregulated in expression. Indeed, about 83 percent of the GOMito DEGs were downregulated in schizophrenia subjects relative to unaffected comparison subjects.
Next we wanted to look at any functional pathway enrichment. To do that, we turned to the ingenuity pathway analysis. We performed that on all of the DEGs, all 356 of those. These DEGs were enriched for three functional pathways that had been given the names mitochondrial dysfunction, OXPHOS, and sirtuin signaling. Anyone who has done these pathway analyses are very familiar with the fact that any one gene that is in one pathway is likely in another pathway as well, especially when they're functionally related. This analysis is no exception.
Here we have the mitochondrial pathway, OXPHOS, and sirtuin signaling. For each one, there is the number of genes, mitochondrial dysfunction is the largest pathway, the numbers of DEGs in that pathway, so 69 percent of the mitochondrial dysfunction genes were differentially expressed, and 96 percent of those were downregulated in subjects with schizophrenia. In our OXPHOS pathway, there were 83 genes, 80 percent of those were differentially expressed, and all of them were downregulated in subjects with schizophrenia.
We see here that 100 percent of these OXPHOS genes are present in the mitochondrial dysfunction pathway. So by and large, the readout of this pathway is really being driven by the OXPHOS pathway genes. In this pathway, these 83 genes are exclusively comprised of the gene components that go towards building those electron transport chain complexes that we saw schematically in the beginning.
Our third pathway that showed up as significantly enriched is sirtuin signaling. Interestingly, just a little over half of these genes overlap with the mitochondrial dysfunction pathway. Sirtuin signaling does a number of things, but particularly it acts as a metabolic sensor, and the pathway genes here really reflect its role in regulating OXPHOS. So we have three functional pathways that are all related to energy production and regulation.
Finally, we wanted to look at the higher-order gene coexpression relationships. To do this, we performed a weighted gene coexpression network analysis in the unaffected control subjects to identify what the relationship between these genes are in individuals unaffected with schizophrenia. This analysis identified five different coexpression modules, identified by different colors. Here is the number of genes within each one of those pathways.
In subjects with schizophrenia, we then looked at the genes for each of those modules, and we see that three modules are enriched for DEGs with the brown module having the most number of DEGs.
Next we wanted to see if any of these modules were enriched for specific functional pathways. What we saw is that indeed the brown module was enriched and it was enriched for the three functional pathways that we identified previously, mitochondrial dysfunction, OXPHOS, and sirtuin signaling, again with the genes being lower in expression.
Finally, in this type of analysis, we wanted to examine whether the fundamental architecture of these modules had been affected by the schizophrenia disease process because they were so enriched with DEGs. So to do that, we performed a Zsummary preservation analysis and the results are shown here with the module size on the x-axis and the Zsummary score on the y-axis. Scores above 10 are strongly preserved and between 2 and 10 are moderately preserved. What we see is that despite the prevalence of differentially expressed genes, all five modules, their fundamental architecture are preserved in schizophrenia.
So to summarize, what we saw is that GOMito DEGs are downregulated in the schizophrenia DLPFC. We saw clustering of GOMito DEGs in functional pathways related to energy production. The WGCNA supports the preservation of the fundamental architecture of GOMito gene expression in schizophrenia. Together what these indicate is a coordinated, downregulation of genes related to mitochondrial energy production.
So what might this tell us about the nature of mitochondrial alterations in the DLPFC in schizophrenia? We see the selective and coordinated downregulation of genes, again, related to energy production. This suggests that there is less ATP being synthesized via OXPHOS. This pattern of alterations, this coordinated downregulation, is consistent with the adjustments that neurons and, really, most cells in your body make, in order to meet lower ATP cellular demand.
Cells regulate the expression of this electron transport chains in a coordinated manner because you need all five of them to work in order to effectively produce ATP. Neuronal ATP demand is primarily driven by action potentials and synaptic signaling. So together, what these suggest to me is that the selective and coordinated downregulation of energy production genes reflects the consequence of persistently lower neuronal firing and synaptic signaling, rather than a frank dysfunction of mitochondrial OXPHOS.
And what's important, or one of the things that may be important about this interpretation, is that this suggests that therapeutics targeted toward enhancing DLPFC circuit activity may benefit cognitive functions and DLPFC cognitive functions in schizophrenia.
So to close, I'd like to acknowledge my colleagues at the University of Pittsburgh and the Translational Neuroscience program, particularly Dr. David Lewis. The funding was from NIMH and BBRF. I would also like to point out and acknowledge the family members who very generously donated brain tissue of their loved ones at a very difficult time in their life in order to be used for scientific research. Without their generosity and their belief in scientific research, we would not be able to do this type of experiment. So I'd like to thank those family members.
And now I'd like to hand it over to Dr. Dost Ongur.
DOST ONGUR: Thanks very much, Dr. Glausier. Good afternoon and good evening to all of you. So I will close out this session by talking about our studies measuring brain bioenergetics in vivo in psychotic disorders. This is my funding and disclosures slide.
So you already heard from Dr. Glausier. She made a great introduction that makes my job a lot easier. Of course, one of the critical functions of the brain is synaptic transmission and information processing. This cartoon, a classic one, shows how critical energy production is for that function. In both neurons on the yellow profiles and the gluteal cells in the blue profiles, the mitochondria, which are these pink organelles, are generating ATP, adenosine triphosphate, which is really critical for maintaining the entire machinery of synaptic transmission.
So much so that some studies have suggested that there is actually a one-to-one stoichiometry between the rate of total neural transmitter cycling and the rate of energy production in the brain so that quite literally every molecule of glutamate being released into a glutamatergic synapse is accompanied by a molecule of glucose being taken up from the bloodstream so that it can be metabolized for energy production.
This kind of research, some of the basic assumptions that have gone into these studies have been questioned. So I don't want to say that this is necessarily the final word in this kind of research, but it's very clear that there's a very close relationship between energy production on the one hand and synaptic function on the other hand.
So of course, if we're interested in neuronal function in psychiatric disorders, then we should also be interested in energy metabolism in the brain in our patient population.
Which brings us to mitochondria. You already heard from Dr. Glausier about the importance of mitochondria for energy production. I'm showing you here a psychiatrist's model of a neuron, this orange circle, very simplified, and inside of it is this massive mitochondrion which is of course where ATP is synthesized de novo and that reaction is catalyzed by ATP synthase enzymes.
But it turns out that as soon as ATP is synthesized in mitochondria, it gets converted to phosphocreatine which is a storage form of the high energy phosphate bond. When neurons actually need new energy, they typically generate that new energy by reversing the phosphocreatine synthesize and creating new ATP. The phosphocreatine storage reaction is catalyzed by the enzyme creatine kinase, or CK. So there is this network of reactions that maintains stable levels of ATP within neurons, but maintain phosphocreatine storage form on the one hand and then the hydrolyzed form which gives us an organic phosphate on the other hand.
It turns out that using a specialized MRI technology called phosphorus magnetic resonance spectrometry, or phosphorus MRS, we can actually measure the levels of all of these important phosphate-containing compounds in the brain, including ATP, phosphocreatine, and inorganic phosphate. In a series of studies along with my colleague, Dr. Fei Du, this is what we've been doing in our patient populations at our hospital.
So one of the first things we did was look at individuals with schizophrenia, chronic patients, along with a sample of age- and sex-matched controls, and we used phosphorus MRS to quantify levels of ATP, PCr, and inorganic phosphate.
So MRS data look like what I'm showing you on the left here. There are actually three residences that correspond to the three phosphate moieties within ATP as well as a PCr residence. These are these peaks in the MRS spectrum.
And of course, the larger these peaks, the higher the concentration of that chemical in the brain. In the very first experiment that we did, we actually found normal levels of high energy phosphate molecules in the brain for both PCr and ATP, and the ratio between the two. People with schizophrenia had essentially normal levels compared to healthy controls. But Dr. Fei Du has been able to implement this ingenious sequence called magnetization transfer phosphorus MRS which actually can look at the dynamic balance between ATP and PCr.
So magnetization transfer phosphorus MRS relies on the fact that there is chemical exchange between the third phosphate, the gamma phosphate, in ATP, and phosphocreatine. This is what the creatine kinase reaction does, it takes the phosphate back and forth between PCr and ATP.
So if you deliver -- and so what I'm showing at the very bottom of the slide is the native spectrum without anything done to it, and you can see the three ATP peaks and also the phosphocreatine peak. Now, if you deliver a radiofrequency pulse that actually scrambles the signal coming from gamma ATP and you continue to collect data as you go up the slide over time, what you see is a reduction in the size of the PCr peak. This is not a real reduction in PCr concentration. It is simply a scrambling of the signal. But of course, the signal gets smaller because molecules of phosphate are being taken from PCr and being brought over to ATP, and then their signal is being scrambled.
So ultimately, the rate of reduction of this PCr peak is the rate of the creatine kinase enzyme. So using this totally noninvasive MRI scan, we can actually quantify the reaction rate of a critical enzyme in the brain, and we have used this magnetization transfer approach to quantify CK reaction rates in the brain in schizophrenia, and we actually see a significant reduction in the exponential reduction of phosphocreatine peak in schizophrenia.
So here I'm showing you at the group level a group of people, healthy controls, healthy comparison subjects, and you can see that their PCr peak is getting smaller over time much faster than that of the schizophrenia group. And when you calculate the time constants for these exponential decay curves, you find that there is about a 20, 22 percent reduction in the rate of new ATP synthesis coming from phosphocreatine in the brain in schizophrenia, and this is a highly significant reduction.
These scans, by the way, are done during rest without the patients conducting any kind of task. So this is just the idling rate of the brain and generating ATP, new ATP, and you'll notice that this finding is nicely consistent with what Dr. Glausier actually just showed us in her postmortem data in terms of an inability to generate new ATP.
Using phosphorus MRS, we can also collect other interesting data, including pH measurements. pH is interesting because of the balance between oxidative phosphorylation on the one hand and glycolysis on the other hand in the brain, when there is a shift from the more efficient oxidative phosphorylation mode of ATP synthesis to the less efficient glycolysis mode of ATP synthesis. Glycolysis is associated with a buildup of lactic acid which reduces brain pH. It turns out we can measure brain pH using phosphorus MRS. I don't have time to go into the details, but we've done that in our patient population, and we see a significant reduction in brain pH in schizophrenia. So this is consistent with a buildup of lactic acid and this shift from oxidative phosphorylation to glycolysis.
In a way, the reduction in pH is the footprint of lactic acid buildup or the downstream consequence of lactic acid buildup, but Dr. Laura Rowland has done similar work, but in her case directly quantifying lactate in the brain, and her group has found exactly what you would expect, that in schizophrenia there's actually an elevation of brain lactate compared to healthy controls.
Since that original study showing reduction in ATP synthesis, we've actually been asking the question what are the functional implications of this reduction in ATP synthesis? So we correlated our phosphorus MRS measures also with functional MRI measures. We collect resting state functional connectivity data from the same patients where we got phosphorus MRS data. So when we place a region of interest seed voxel in the medial prefrontal cortex, we can identify the default mode network, the classic DMN, in the medial PFC, the retrosplenial cortex in the lateral parietal regions, as well as an anticorrelated task-related network in the insula and then the prefrontal cortex, and then the lateral parietal cortex.
We have done this in our own patient population as well as healthy controls, and when we asked the question what is the relationship between energy production on the one hand and functional connectivity on the other hand? Here is what we see. In the left panel, I'm showing you the within-network connectivity in the DMN, and on the right panel, I'm showing you the anticorrelation between the DMN and the task-related network. In both cases, green is healthy controls, and you can see a strong relationship between the energy supply on the x-axis and functional connectivity on the y-axis that's positive in the case of within-network connectivity and negative in the case of between-network anticorrelation.
Interestingly, this relationship is actually broken down in people with bipolar disorder in blue and people with schizophrenia in red, in both the within network and the anticorrelation between network anticorrelation. So this suggests that not only is there a reduction in the availability of ATP that's being generated, there's also a breakdown in the ability of that new ATP to contribute to long distance connectivity in the brain in these largescale neuronal networks.
There's some suggestion that the longer-range connectivity is actually more impaired as a result of the energy supply issues as opposed to local connectivity that we also discussed in this paper.
We also are curious about the relationship between these brain abnormalities and peripheral abnormalities. So in this case, we actually studied along with Virginie-Anne Chouinard, insulin sensitivity in a group of healthy controls in a patient sample as well as siblings of our patients. These are generally medically and psychiatrically healthy younger people, close in age to our patient population in the study, and they all underwent the glucose tolerance test, which allowed us to quantify insulin sensitivity with good fidelity, and what we found was this amazing reduction, about a 50 percent reduction in insulin sensitivity in people with psychotic disorders, but also their first degree relatives who are otherwise healthy, lean, with no prediabetes or diabetes, and yet, their insulin sensitivity is actually quite impaired.
In a small sample, we also see the relationship between insulin sensitivity on the one hand and the matrix composite total score on the other hand, suggesting that this peripheral insulin sensitivity index may also have implications for brain function as we would have expected.
You also saw a version of this electron transport chain cartoon from Dr. Glausier, and here I'm showing the electron transport chain that takes place in mitochondria. Because of our ability to quantify NAD and NADH, this is a redox pair and the balance between NAD-plus on the one hand and NADH on the other actually gives us information about the predominance of oxidative versus reductive reactions in the brain, and we're also able to quantify NAD and NADH in the brain using phosphorus MRS. NAD biology is very interesting, very significant. It's not only involved in mitochondrial function and metabolism and redox reactions but also in circadian rhythms, inflammation, and several other important processes. So having the ability to quantify NAD in the brain is actually potentially of importance for future studies.
When we have looked at the NAD/NADH ratio, sometimes called the redox ratio in the brain, we see a significant reduction in people with chronic schizophrenia. This is actually also there in people with first-episode schizophrenia and people with first episode mania and psychosis are somewhere in between.
Interestingly, this NAD/NADH ratio has a very strong age dependence. In healthy controls, you can see the redox ratio really peaks in your mid- to late-20s, and then there's a reduction until you go into your 50s, whereas in people with schizophrenia, this age dependence is weaker but primarily because it was never very high to begin with.
The redox ratio abnormality in this patient population actually is not in the oxidative stress direction but in the reductive stress direction in our case, which is an interesting finding, and I think we really need to do a lot more work to understand the relationship between oxidative and reductive stress in the brain in schizophrenia. There's some recent interesting work on this from Dr. Kim Do and her group in Lausanne, Switzerland, and I think really we need to do a lot more work on our redox abnormalities in this population.
This slide puts together a lot of research that we've done over the past decade. Each of the circles that I'm showing you in our first episode and chronic patient populations are our own data. I didn't have time to go through all of them, but what I'd like to show here is the complexity of various measures.
In the case of glutamine, you see this biphasic abnormality where glutamate levels are elevated early on in the brain, and then reduced later on. In the case of redox dysregulation, it's more pronounced in first episode, and there's a little bit of an amelioration in chronic patients. And then the CK reaction rate reduction actually is more progressive as you go from early to late stage disease.
So again, in the interest of time, I'm not going to go into detail for all of this, but I'm going to leave you with this one potential interpretation that's consistent with the data. It's possible that there's a cascade of abnormalities that starts with elevated circuit activity in the earliest illness stages. That leads to biochemical distress in neurons in schizophrenia and the redox imbalance, which actually creates this compensatory slowdown in ATP generation and shift from OXPHOS to glycolysis, which then feeds back to downregulated circuit activity that we see in the case of the connectivity abnormalities in chronic patients.
But this actually gives us a chance to normalize the redox imbalance, relatively speaking, with the loss of long distance circuit functions. So this sort of iterative relationship between biochemical and circuit level activities, we think, are feeding into one another as the illness progresses from early stages to later stages.
I'm going to close by thanking all of my colleagues, especially Dr. Fei Du, who has implemented the MRS sequences and worked on the phosphorus MRS studies with me over the last decade, and several of my other colleagues, and also all of our patients and their families who volunteered for our studies. Let me leave you with an aerial view of McLean Hospital in the suburbs just outside of Boston.
I'm going to turn it back to Dr. Judy Ford now.
JUDY FORD: Thank you, everybody, for four wonderful talks. They are very exciting and I'm looking forward to the discussion amongst all of you, as well as the other panelists. So all of you have turned on your cameras. That's great. I'm going to follow the good patterns of my predecessors and start with questions for the first speaker, which would be Dr. Baran.
Dr. Baran, thank you. That was an excellent exciting talk. I was really fascinated by how you conditionalize the white or the pink noise stimulation based on where a person's -- I guess it was slow wave oscillation, cortical oscillation was. That sounds like a really great way to try to increase the spindle density, and I think I know the answer to this, but I just wanted to ask you anyway. If you can treat spindle density, like I think you managed to with the eszopiclone, I'm not sure how to say it, or even with this pink noise, would you expect there to be a change in the corticothalamic connectivity? I know that's a more static kind of measure, but on a long-term basis, do you think that you might actually get at some of that function with your methods? Because I know from the literature that in fact if you do have a good function of this thalamocortical connectivity, in fact you might have fewer positive symptoms, fewer negative symptoms, et cetera.
BENGI BARAN: That's an excellent question. I view this thalamocortical hyperconnectivity as a trait level alteration, and we have evidence from genetic studies, for instance, that gives us the impression that this is a trait level alteration. So I would not expect that. If anything, boosting sleep spindles would be trying to overcome this abnormal hyperconnectivity.
JUDY FORD: Okay, thank you. Very interesting work. Congratulations on that.
Let me ask a question then of Dr. Lee. Dr. Lee, there's one question that was from the audience that said patients -- so are we to surmise that patients will underreport poor sleep, and how can we better address sleep quality clinically?
ELLEN LEE: I think this is a really good question, because there are so many layers to this. I think that we found in many other studies that subjective sleep complaints do not always align with objective sleep complaints, and I think that there's a lot of influence of mood symptoms and other issues that may influence that, but yet subjective sleep quality is an important metric that is related to a lot of important outcomes.
I think clinically my sense is that we tend to overlook sleep problems in people's psychotic disorders. We tend to attribute them to the main symptomology of the disease and we tend to treat it by just providing a slightly more sedating antipsychotic or slightly more sedating antidepressant to see if that will work, and I think sometimes we don't look for a primary sleep disorder as well, and I think that's a missed opportunity clinically.
JUDY FORD: Great, thank you. Next would be Jill, Dr. Glausier, and this is a question, it says are any of the mitochondria genes in your studies part of the WAS genes associated with schizophrenia?
JILL GLAUSIER: Yeah, thank you for that question. A lot of the work that has been culminating and coming out from the schizophrenia genomics workgroup has really, their GWAS studies, have really started to point toward synaptic genes rather than genes traditionally associated with mitochondrial functioning, and in fact, some really great studies that have been done specifically looking at genes related to mitochondrial functioning, Mark Fauder(ph.) has done a lot of these, have not found huge effects or strong associations with schizophrenia.
So the genetic GWAS studies are really starting to point towards what we might consider idiopathic schizophrenia, a bunch of alterations, mostly in synaptic-related genes. So this is another point of evidence that starts to converge on the idea that synaptic dysfunction will cause lots of different effects in the brain in order to -- you know, your brain is always trying to keep itself working and going the best it can, and so that we might want to look at some synaptic alterations. But great question.
JUDY FORD: Let me just follow on that with a question that occurred to me. I think you mentioned that perhaps this would be a treatment target, and I wonder whether you think something like biofeedback training or TMS or something that might increase the neural activity. I think you were saying that neural activity is what's deficient in the DLPFC and possibly it could be treated, but I'm wondering what kind of treatments you think might be possible and relevant.
JILL GLAUSIER: Yes, so Dr. Ongur brought up a good point that there's relatively good evidence, and Dr. Rowland has I believe been part of this, as well, that very early in the disease there might be increased activation or increased activity within the cortex, but relatively quickly one might say it comes to this sort of hypofunctional or hypo-glutamatergic state within the cortex, and the types of analyses I'm doing in postmortem human tissues is looking at individuals who are not in first episode.
So treatment opportunities for this, I think, would, for individuals who are not first episode, would absolutely include things like TMS, and there are a lot of treatment paradigms that are being designed and delivered right now in experimental ways for TMS and schizophrenia. Sometimes trying to improve psychotic symptoms, but other times trying to improve cognitive functioning as well. So it is, I think, what my data is telling me is that if you were able to get the neurons to fire more, that the mitochondria could stand up and they could do it.
JUDY FORD: I want to follow up on that in a minute but let me ask Dost a question that was posed. This is a question from Robert Green. He says in South Carolina we have some patients diagnosed with schizoaffective disorder who rapidly develop encephalopathy during carnitine depletion. Could this be mitochondria based?
DOST ONGUR: Carnitine is a small molecule that's an important cofactor for multiple bioenergetic reactions in mitochondria. There is a small literature on carnitine supplementation, L-carnitine is actually -- you can get it in GNC. Athletes sometimes use it. So it plays these biochemical roles.
I think the broad scoping answer that I would give is that people with psychotic disorders -- schizophrenia, schizoaffective disorder -- are more prone to biochemical distress in brain function. So I'm not surprised to hear that they might react negatively. I don't have firsthand experience with carnitine depletion, but they might react negatively compared to people who don't have schizophrenia.
This also reminds me of two things. One is there's an old literature suggesting that people with schizophrenia actually, a brief glucose load improves cognitive function. In the 1980s and 1990s, there were several papers suggesting this. So it's unfortunately not a long-term strategy. Hyperglycemia is not a good thing in the long run, but a brief glucose load actually improves cognitive function in people with schizophrenia.
But then on the other side, there is growing interest in ketogenic diets and low carb diets in schizophrenia, because they might actually sort of force to upregulate mitochondrial function by depriving mitochondria of the easy fuel, which is glucose, and forcing mitochondria actually to become more efficient.
So the biochemical distress probably goes in both directions, but might actually be useful for improving brain function.
JUDY FORD: I am so glad you mentioned ketones. This is my new favorite cure-all for every problem people with schizophrenia have. Not only do we know that glucose metabolism in the brain is problematic in people with schizophrenia, but ketones might rescue that for them. And also, a byproduct would be possibly they could lose some weight and become -- in fact a lot of the inflammatory markers would probably decline.
DOST ONGUR: Absolutely. You know, diet-based interventions may be more acceptable to a broader range of people compared to medications. So there are several good things about them, although the ketogenic diet is not easy to do. So if anybody has tried it, it's difficult to sustain. But it's definitely something -- there's growing interest in it.
JUDY FORD: I keep almost trying it. I have a big magazine I got myself for Christmas called Keto Diets. I actually do have a clinical trial starting this month to treat people with schizophrenia with a ketogenic diet. So stay tuned. It's basically free food, they get free food. I don't know if that's going to be an incentive or not, because it's not necessarily the food they want to eat.
I think there are probably some other questions before we move to the panel discussion that I wanted to ask. I know, Ellen, do you think that people with -- in fact someone asked you this question. Maybe this is for the panel discussion, but I was going to ask it anyway, and that is: are there any things that you can think of that will make the CPAP and even tests of sleep apnea more acceptable to people with schizophrenia?
ELLEN LEE: This is a great question, and I think part of my interest was because of some of the innovations we've had in sleep medicine. The home sleep test is in many ways an easier thing for many people to do. They don't have to come into the lab. They can do it at home. There is no EEG component, unfortunately. So we don't gather that data and so it cannot provide everything we need. It does require that patient to put it on themselves. However, it's a lot easier to administer. They can try it for multiple nights if it doesn't work the first night.
In terms of making PAP treatment more acceptable, I think again there've been some innovations here where we've changed the types of masks that we provide people. There's more ability to humidify. You can do automated settings that can be more comfortable to adjust the pressures. But the other part of it is there may be pharmacotherapies that can help with the arousal threshold and the sort of the reaction to hypoxia, your body's reflex to hypoxia, that may make it more acceptable. So there are studies actually going on at UCSD led by Dr. Malhotra here, looking at how different medications like eszopiclone, for one, as well as acetazolamide could make PAP treatment more acceptable or even replace PAP treatment for certain types of patients. There's a lot of interesting stuff on endophenotypes of sleep apnea.
JUDY FORD: Thank you. We now need to move into the panel discussion and so we need to invite Dr. Frangou to join us. There she is. And Ellen, Dr. Lee, maybe you want to kick off the panel discussion with any questions or comments you have for the session?
ELLEN LEE: Well, I think these are really great and interesting talks, and I'm really excited to see how they can wrap into some of the other talks we've already had with clinical symptomology and cognitive functioning, and I was wondering if the other panelists could talk a little more about how they think their findings would be relevant for future interventions and future strategies to study some of what you study in older patients with schizophrenia.
JUDY FORD: Who wants to take that?
DOST ONGUR: Maybe I'll take a crack at it. I think the way I've started thinking about sort of the evolution of brain abnormalities in schizophrenia is that it was said earlier today in one of the earlier sessions that somebody who's been living with schizophrenia for some time has a brain where a lot of things have happened, and you have a different situation. It's not the same brain that you have at the first episode or in an at-risk state. So there's a bit of archaeology that needs to be done. Clearing away the layers of compensatory changes and all the feedback and feedforward circuits that have taken place, and that's really the process that we need to understand to understand aging and later life in schizophrenia, and I think -- today's meeting really is sort of good evidence that this is starting to happen, but I feel that the field generally has not spent a lot of time thinking about this active evolution of abnormalities. There has been a lot of searches of the lesion. You know, what is the original problem and how does that lead to schizophrenia, and that's not really how this unfolds over time. So it's a big picture. A more vague answer, but I think it's an important conceptual leap.
BENGI BARAN: I can take a second crack at it, if that's okay, and that's going to be a non-answer obviously. So even for somebody with a very narrow interest, the cognitive role of sleep, and somebody who has studied both aging and psychotic disorders, these have always been compartmentalized in the kind of work that I've done, and there was a question in the chat box asking if spindle deficits are worse in older patients versus younger patients, and the answer is we don't know. We haven't looked at it. If anything, the developmental trajectory of sleep oscillations is unfortunately largely understudied.
My current interest is on the much earlier side of things, whether changes from childhood to adolescence to young adulthood, if there is, if something goes awry in that trajectory to predict psychosis or predict clinical risk of psychosis. But none of us have looked at the other end of the spectrum, if aging produces additive cognitive deficits in schizophrenia patients.
If anything, in the chronic schizophrenia data that I presented, we limited us to a certain age range so as not to confound aging effects with the disease process.
JUDY FORD: Thank you, Bengi. I have long noticed ever since early on doing event-related potential research with people with schizophrenia what we called accelerated aging in schizophrenia, and it makes me wonder whether some of the things we see even early in the illness course, maybe in the 40s, are what a healthy normal person might look like at age 60 on some of the scales that we look at. And again, this reminds me of the question I think that Dilip addressed earlier, how do you tell the difference between psychosis as a result of dementia and dementia as a result of psychosis? It may lay in the patterns of deficits that we see.
There's another question in the chat that was addressed to Ellen, and this is from actually Dr. Baran. You could ask it yourself if you would like, about interventions that target weight management.
BENGI BARAN: This just gets at the effect, compounding effect of BMI.
ELLEN LEE: I think you're absolutely right. Like one of the sleep researchers I work for, says the only surgery that works for sleep apnea, that's most effective, is bariatric surgery. So we know that weight loss and improving obesity is a huge risk factor for sleep apnea, and that is not different in people with schizophrenia.
What's interesting though is there have been studies, many from younger patients with, for instance, posttraumatic stress disorder, where we see people who do not have the risk factor of obesity, yet they have high incidence of untreated sleep apnea, and that seems to be related to actually their physiological arousal overnight, and so there maybe some aspects of that happening in schizophrenia as well. We know there is a high rate of trauma in this population. So that's another consideration is when there are brain factors, hypervigilance, that may be contributing to sleep apnea.
But I want to make sure Sophia has her chance, too.
SOPHIA FRANGOU: No, I was just, as you were talking, I was just thinking that a lot of the arousal systems, and perhaps we will hear from Anissa tomorrow, are regulated by cholinergic mechanisms, and cholinergic mechanisms are also critical during development, including intrauterine development, in actually putting the brain together. I was wondering whether what we are seeing are different aspects of miswiring, that, depending on the degree perhaps and other personal characteristics, it may affect arousal systems like breathing and sleeping more than other systems. I was kind of wondering whether this is again another consideration that they are not -- that it's part of, an inherent part of the disease, not necessarily totally attributable to other external risk factors like weight gain following antipsychotics or lifestyle or whatever. That's simply what came to mind. But I don't know if anybody thinks this is a worthwhile avenue of considering cholinergic arousal mechanisms other than obesity.
ELLEN LEE: Right. I think they are going to start with the lowest hanging fruit, but we are looking at some of this with a large database of people who have had sleep studies, like PSG sleep studies done at UCSD, and we are seeing more of the mixed central and obstructive picture in our small group of people with schizophrenia, compared to the general population. So we do think there might be some differences in their phenotypes. So we're trying to expand our database to try to get to what are the features of sleep apnea that are unique.
SOPHIA FRANGOU: My other comment, if I'm allowed one more, is to try to link perhaps our first session with our last session. In the first session, we talked a lot about sex and sex differences. Is there anything in terms of the metabolic energy pathways that can sort of tell us anything about sex differences in age of onset or any other aspect of psychosis? I presume the short answer is we don't know that much yet about it.
JILL GLAUSIER: I will say that, talking about postmortem studies for schizophrenia, we have in the level of resolution of analysis that we're doing, we have not seen huge sex differences in the types of dependent measures that we look at in schizophrenia. That's different, however. I know this is the nonaffective psychosis workgroup. But that doesn't translate to other disorders. So we do see pretty big sex differences, for example, when we're looking at major depressive disorder in males versus females.
And I also want to bring up that this can also be brain region specific. So for example, a lot of my work has been in the DLPFC and other cortical regions, but perhaps in subcortical regions, that there are sex differences that are more pronounced and that it is brain region specific. So we have not seen huge sex differences for schizophrenia, but we absolutely have seen them for things like major depressive disorder.
SOPHIA FRANGOU: And in the imaging data, there hasn't been a dramatic effect of sex, has there?
DOST ONGUR: It is also not well balanced. So we have fewer women in all of our, in each of our studies, and when we have looked at it, we haven't seen anything, but it's not surprising, because it's so underpowered.
JUDY FORD: Are any of the effects that people have been talking about affected by medication? You know, thinking about Larry Yang's wonderful talk and treatment-naïve people. Do you have any information about whether medication has any effect? I do know that before antipsychotic medications even came on the scene, people with schizophrenia had type 2 diabetes, they had insulin-resistance, et cetera. So those kinds of things I don't think are directly related to medications. But what about some of the other variables that you guys are looking at?
DOST ONGUR: I can address to that a little bit for our work. For in vivo neuroimaging, this is of course a big concern. People have the medications on board.
We have a few lines of evidence that suggest that medications are not a big factor here. We can never really fully rule them out. For all of our studies, we look at chlorpromazine equivalents, and we've never seen a relationship there. For several of our studies, we have subgroups of patients who are choosing not to take medication and they volunteered for the research. Typically, if you have an n of 25, 26, you have 5 or 6 people who are actually not on medication. They're not medication-naive, but they're not on medication when they volunteer for the research, and their results are always squarely in the middle of the patient group. So they don't seem to stand out in any way.
And then I showed a little bit of the relatives data that suggests for various of our measures there's shared -- vulnerability to illness appears to be linked to the results, and not the medication. And then one thing we have some unpublished data that we're working on now; we just finished a study looking at olanzapine treatment in healthy individuals. The IRB allowed us to give people five milligrams of olanzapine a day for two weeks, which is not a full dose, but it's something, and we have scans before and after and we see no change. So for low dose olanzapine, it suggests that some of these measures are not being impacted. It's reassuring that these aren't all medication effects.
JUDY FORD: Yeah, the sibling data I think are very strong, assuming those siblings were not on medications of some kind. They could have been. Are there questions among the panelists? I'm trying to keep track of all of them. I don't know that we've addressed all of the questions that you had for each other.
Let's see. I think Dr. Ongur had a question for Dr. Glausier.
DOST ONGUR: Yes. You addressed this a little bit in terms of the activity dependence of the transcription abnormalities, but I was wondering whether there is any background knowledge about some of these genes being more sensitive to activity levels and can you leverage that. If those were the genes that are more abnormal in schizophrenia, that suggests that the underlying mechanism is in fact activity dependent. Is that possible?
JILL GLAUSIER: Yes. You are familiar and I'm sure everybody else is, there's oxidative phosphorylation as we presented, you have to go through electron transport chain, but there is brain bioenergetics that you presented, so it's a huge intermesh of a lot of different functions that have to come together in order for oxidative phosphorylation to work and produce energy effectively. So all of those pathways and processes have not been rigorously examined in experimental models.
What has been pretty rigorously examined are the expression of the genes that make up the electron transport chain. Both those genes that are encoded by the nucleus, which is what I presented today, and the handful of genes that are produced by the mitochondrial DNA, which I did not show data for today, and there's some technical reasons as to why quantifying the mitochondrial-generated ones in the data I presented today might not be coherent with the type of technique that we use in doing nuclear encoded is more interpretable.
But both nuclear and mitochondrial-encoded genes for the electron transport chain have been rigorously examined, both in vitro and in vivo, how they change in response to different types of perturbations to chronic changes in neuronal firing. So if you just do a short chain, so let's say you just put a little bit of TTX on or something that deprives the cell for a little bit of time of firing, actually the transcription and translation of these goes up.
But if you chronically reduce activity, they all go down, and the nuclear and the mitochondrial encoded ones do, and they do it consistently and coherently together. This has been shown for example in cell culture with TTX. This has been shown with monocular deprivation in in vivo animals, in both monkeys and rats. So Margaret Wong-Riley did a lot of this work. So those set of genes, the ones that we saw were most changed are the ones that have been best studied and most closely linked to changes in neuronal activity.
JUDY FORD: Jill, I have a question, too. You said that the downregulation of energy production genes reflects a consequence of persistently lower neuronal firing in synaptic signaling, rather than frank mitochondrial oxidative phosphorylation. Do you therefore think that my idea of a ketogenic diet is not going to hit the target that I am hoping it will hit?
JILL GLAUSIER: I think this raises an important point, because there's still a lot of kind of mysteries about the ketogenic diet and its effects and what types of cells can be affected in specific ways in the brain. But I think to me the important part to answer that question is that there are many etiological pathways to a diagnosis of something that is in schizophrenia spectrum. What I presented today and most people have presented today are means and the average differences between groups of people. The subjects that I included are individuals that I would say have sort of idiopathic schizophrenia. I have no idea what caused it.
And I can't say that that necessarily means that the overall pattern of changes that I see in this large group of people completely reflects any given individual that I pull out. Schizophrenia has an incredibly heterogeneous clinical presentation, and there's many different ways to get to the same effect on a neuron. So there could be a subgroup of individuals and maybe it's the larger of the group that it is the synaptic mechanism.
But there might be a subgroup of individuals that have a mitochondrial etiology and it's the minority of the group, and maybe there's something specific or special about those subjects that we just haven't identified yet.
And then of course, that's totally different than individuals who have been diagnosed with a schizophrenia spectrum disorder for which we do have an idea of what the etiology might be. So 22q, NMDA encephalopathy, those types of things. So I think it's really important to keep in mind the heterogeneity of the disorder that seems to be present at literally every level of analysis and perspective of schizophrenia.
JUDY FORD: Thank you. With that, I am afraid we have run out of time. I want to thank everybody for a very exciting discussion and very exciting data and presenting it, sticking with us through all of this, and as we search for treatments and cures, and I'm very encouraged by what I'm hearing today.
So thank you very much. Please tune in tomorrow, same time, same place. See you then.