Skip to main content

Transforming the understanding
and treatment of mental illnesses.

Celebrating 75 Years! Learn More >>

Workshop: Nonaffective Psychosis in Midlife and Beyond - Day 2


ELLEN LEE: Good morning, everyone. Welcome back to the second day of the NIMH Virtual Workshop on Nonaffective Psychosis in Midlife and Beyond. I'm Ellen Lee. I'm co-chairing this session with Dr. Sophia Frangou. I would like to also thank our NIMH organizing committee of Drs. Laura Rowland, Andrea Wijtenburg, and Craig Fisher.

So first I'm going to start by just going over a little bit of housekeeping issues for today. Attendees are entered into the workshop as listen-only mode with cameras disabled. If you would like more information on the speakers for today, we invite you to read their biographies which are on the event registration website.

If you have any questions, please submit them through the Q&A box at any time during the presentation. You can also be specific about who you'd like to address your question to if there's a specific speaker you'd like to respond. We will have a Q&A session at the end of all of the four speakers for each session followed by a short panel discussion with all of the speakers and the chairs.

If you're having any technical difficulties in hearing or viewing the workshop, you can note this in the Q&A box, or you can also send an email as listed here. This workshop will be recorded and posted to the NIMH website for later viewing.

Just to give you a quick overview of what we're expecting to talk about today, we're on day two, so we'll have sessions on accelerated aging, treatment targets and intervention development, as well as a discussion on the synthesis of the topics covered today, opportunities, and next steps.

I'm going to give a quick recap over some of the sessions we had yesterday. Session one was on phenomenology, course, and outcomes, and we started out with a wonderful discussion by Dr. Carl Cohen who provided some excellent insights into the heterogeneity of recovery and remission outcomes in longitudinal studies of older adults with schizophrenia.

Dr. David Castle then discussed the striking differences between risk factors and presentation of early onset and late onset schizophrenia for very large epidemiological studies. Dr. Anthony Ahmed provided a deep dive into the understanding of negative symptoms across the lifespan as well as their effects on cognitive functioning.

Then Dr. Iris Sommer gave a fascinating talk on the sex differences in psychosis, covering both the neurobiological differences as well as pragmatic concerns in designing future research studies to improve clinical and treatment outcomes for women with schizophrenia.

Then there was a Q&A and a panel discussion which covered many topics from the definition of the phenomenology as well as understanding how to better serve these populations.

Now I'm going to hand it over to my co-chair, Dr. Sophia Frangou. Thank you.

SOPHIA FRANGOU: Thank you, Ellen. Continuing our recap of what was discussed yesterday, in session two, we focused a little bit more on how to capture cognition in schizophrenia across the lifespan.

We started with a presentation by Dr. Kotov that highlighted the fact that there have been multiple epidemiological studies suggesting that schizophrenia is associated with a much higher incidence of dementia, especially later in life. Then he focused on the cohort that they have been following up for nearly 25 years within the Suffolk County where they very clearly show accelerated decline in cognition associated with significant decrease in terms of their overall patient function as patients become older and this acceleration seems to become more pronounced towards the later part of life.

The second presentation by Eva Velthorst focused more specifically on social and emotional processing and combined findings from adult study and also findings from the Suffolk County sample. So in terms of emotional processing, at least in the data that was available, there doesn't seem to be a very marked decline as we see in the sort of non-active cognition, but we see more or less stable deficits in patients, but as they go more to the older age differences between patients and controls are less pronounced in kind of neutral stimuli and fearful stimuli, but remain quite pronounced in our processing of happy faces.

Dr. Lawrence Yang then described to us a really amazing study that is happening at the moment in rural areas in China. The investigators have access to a population living in a quite remote area that has a very unique population of patients that have never received treatment for schizophrenia because of a combination of lack of motivation to do so and lack of cognitive ability. So as part of this study, they are offered the opportunity to receive treatment if they want to, but at inception the investigators have access to an untreated sample of patients with schizophrenia.

Here, of course, the duration of illness and duration of untreated psychosis is exactly the same thing. They find what has been shown in many other studies, that the duration of untreated illness is negatively associated with cognition. As part of the same study, there are planned analyses to compare these never treated patients to patients that have received antipsychotic medication to clarify the contribution of antipsychotics in positive or negative in terms of cognitive trajectories.

The final talk on session two was by Dr. Ivleva. She focused specifically on longitudinal changes in the volume and function of the hippocampus, showing that there is a sort of anterior to posterior spread in terms of dysfunction and a hyperfunction of the early stages of the illness followed by hypofunction later on.

She also highlighted a number of other neuroimaging and peripheral biomarkers that seem to have complex associations with duration of illness. She emphasized the importance of having a multimodal and multi-panel assessment of patients when we try to capture the association between age and different markers of disease progression or disease expression.

Session three started with a presentation by Dr. Baran, who reminded us that the density of sleep spindles is very closely linked to thalamocortical activity and memory consolidation in the general population, and that the density of sleep spindles in patients with schizophrenia is actually quite reduced, which may contribute to inabilities in verbal learning and general learning that have been very well documented.

She highlighted the fact that just addressing the density issue with pharmacological manipulation is not sufficient, because this regulatory system is actually quite complex and any potential treatment intervention has to incorporate multiple other targets because the density of the sleep architecture and the general oscillatory activity within different networks are quite interconnected, so one cannot just change one parameter without considering how it will affect the entire system.

Dr. Lee highlighted the fact that obstructive sleep apnea is an unrecognized and potentially very treatable contributor to dysfunction in schizophrenia, both in terms of (indiscernible) but also in terms of depression and other symptoms, and suggested ways that we can improve the way that we measure this in a fashion that is easy to implement by patients at home and goes beyond just subjective ratings of sleep apnea. We also discussed the pathways that may be associated as a cause or as a consequence of sleep apnea in these patients that involve systems that regulate oxidative stress and systems that regulate arousal.

So this talk linked really very nicely to the subsequent two talks. Dr. Glausier used her transcriptomic work to draw attention to the fact that gene expression within pathways that are related to oxidative stress but also to energy generation, specifically within the prefrontal cortex, seem to be abnormal in schizophrenia in a way that suggests the inability of patients with schizophrenia to access ATP from its stored form, thus perhaps shifting the way that the brain generates energy to a more glycolytic pathway that may not be optimal.

This was further reenforced by the last presentation by Dr. Ongur that used neuroimaging and introduced us specifically to an innovation by Dr. Fei Du in terms of capturing the energy signal and the pH signal of the brain using phosphorus MRS, where they also found that there is a problem with patients' ability to generate energy, and that is problem is directly, perhaps, linked with difficulties that we see in neurotransmitter function because the two phenomena are quite linked.

We talked a little bit more in this session about whether sex differences that we see the prevalence of later onset schizophrenia can be associated with changes of other neuroimaging signals or transcriptomic profiles. The general agreement is that if there is a sex difference, it doesn't seem to be very pronounced, and it has not been sufficiently captured by studies so far.

So that ended the first day of our workshop. As Ellen said, it was a very lively, dynamic, and very interesting discussion that followed each of the sessions. We hope the same for today.

But before we start with our sessions, I would like to introduce Dr. Lisanby, who is going to address our panel and share her thoughts with regards to the importance of looking at nonaffective psychosis later in life and perhaps introduce some of her thoughts in terms of how this could be incorporated within any neuromodulation perhaps approaches to improve cognition generally in schizophrenia and perhaps later in life.

Holly, up to you.

HOLLY LISANBY: Great. Thank you. Just listening to that fantastic recap of yesterday really tells us that we're in store for a real exciting day today.

So it's my distinct pleasure to welcome you to day two of the NIMH Virtual Workshop on Nonaffective Psychosis in Midlife and Beyond. My name is Dr. Holly Lisanby, and I direct the Division of Translational Research at NIMH. We are really delighted to be sponsoring this workshop.

I want to start by thanking our workshop co-chairs, Drs. Ellen Lee and Sophia Frangou, our NIMH planning committee, Dr. Laura Rowland, Dr. Andrea Wijtenburg, and Dr. Craig Fisher, our many illustrious speakers, panelists, and moderators, and of course, I want to thank each of you who have chosen to spend these two days with us to focus on the important public health topic of nonaffective psychosis in mid- to late-life.

I'd like to say a few words about why this is a really important topic for us, because at NIMH we are committed to accelerating the translation of discovery and recovery from mental illness across the lifespan. While we often think of schizophrenia as a neurodevelopmental disorder that primarily has its onset in adolescence and young adulthood, it turns out that a sizeable proportion of individuals have their onset later in life. Understanding these differences can be really important to making a difference in these conditions.

As a geriatric psychiatrist myself, I am especially delighted that NIMH is hosting this workshop which addresses first onset psychosis in later adulthood, a less commonly researched topic, but very important.

Mid- to late-life is a period of a lifespan when psychotic spectrum disorders can be particularly burdensome and disabling. So it's vital that we find effective interventions to enable older adults living with psychosis to fulfill their potential and to thrive through midlife and beyond. The experimental therapeutics approach that we have adopted at NIMH to developing novel interventions involves discovering and validating treatment targets, demonstrating that the therapeutic intervention engages the target, and testing whether, when the target is engaged, the condition improves.

Given the complex confluence of neurodevelopmental processes, neurodegenerative processes, aging processes impacting brain and body, and the accumulation of exposures that individuals experience in mid to late life, we may not expect the therapeutic targets for later onset psychosis to be exactly the same as those seen in early onset psychosis. In addition, individuals who develop the onset of psychosis early in life experience changing needs as they grow and develop into older adults.

So understanding how to best optimize functioning and support healthy aging for older persons living with psychosis is a priority for us as well.

So we've convened this workshop to identify the state of the knowledge of this field, to define promising opportunities that have the potential to advance scientific understanding and stimulate intervention development, and to shed light on the areas where research could make the most impact in the lives of people living with nonaffective psychosis in mid to late life.

So to kick off today, I'd like to share some of my personal highlights from day one. You've already heard a fantastic recap, but these were some of my take-homes. Of course, I have to start with Dr. Roberta Payne who poignantly shared with us her lived experience of schizophrenia and periods of her suffering, which she called her mud years, where she felt as if she were encased in mud and couldn't break out of it.

She shared with us about the medications, the psychotherapy, the stress management, and mentorship that helped her restore her functioning. She shared how living on a fixed income as an older person and having to deal with health issues that come in later life have impacted her.

From our other speakers, we learned that the phenomenology and course of psychosis in late life is very heterogeneous and can differ between men and women, which argues for a more nuanced phenotypic understanding that moves beyond categorical DSM diagnoses and highlights the need for large sample sizes with harmonized measures.

We learned that to understand the drivers of cognitive decline in older adults with psychosis, we need to be able to disentangle the various contributions of aging, cerebrovascular disease, antipsychotic medication exposure, and sleep dysfunction. We learned how biomarkers such as cerebral blood volume, connectivity measures, and MR spectroscopy can help us begin to understand these complex processes and their trajectories over time, and ultimately may enable us to parse markers of aging from markers of disease progression. We learned about how human postmortem work and in vivo imaging can shed light on the role of brain energetics and mitochondrial function.

So today, we look forward to another set of exciting speakers who will take us deeper into the topics of accelerated aging, premature morbidity and mortality, treatment targets and intervention development, synthesis opportunities, and next steps. I look forward to spending the rest of the day with you and with our wonderful speakers and learning from this outstanding group of scientists who have dedicated their careers and their research efforts to helping us understand this important public health issue.

We deeply appreciate your input into where research can make a difference in the lives of individuals, families, and communities impacted by nonaffective psychosis.

So now, without further ado, I'm going to go ahead and pass the virtual microphone over to Dr. Dilip Jeste who will moderate our first session of the day on the important topic of accelerated aging. Dilip, please take it away.

DILIP JESTE: Thank you very much, Holly. Good morning, good afternoon to everyone. I again want to thank Laura Rowland and Holly Lisanby and other NIMH leadership for putting together this outstanding webinar and inviting me to be a moderator.

I also want to compliment the co-chairs, Ellen Lee and Sophia Frangou. They did a really fabulous job yesterday and I'm sure today, too. It's a lot of work with five sessions and 30 speakers.

So I am delighted to welcome you to this session on accelerated aging. We will have four speakers. Each speaker will talk for exactly 15 minutes. After that, we will have ten minutes of Q&A. Any questions you have, please put them in the Q&A box, and you can do that while people are speaking. Please identify what speaker the question is addressed to unless it is addressed to everybody. After that, we will have a 20-minute session for panel discussion in which the group co-chairs will join.

So let's get started. The first presenter is my colleague, Dr. Lisa Eyler from University of California, San Diego. Lisa?

LISA EYLER: Thank you so much, Dilip. I'm delighted to be here today to talk to you about systemic biomarkers of advanced and accelerated aging. I'm going to be presenting work that was done primarily with my colleagues at UCSD and also at the VA MIRECC in San Diego. A lot of the work will have been led and spearheaded by our moderator, Dr. Dilip Jeste, and a lot of it is also work done by one of our co-chairs, Dr. Ellen Lee.

So I wanted to start with a general question about what do we mean by accelerated aging? This topic came up a bit yesterday and there were a lot of different graphs that were shown about the trajectories across the lifespan of cognition and other variables. I thought that this schematic might be helpful to distinguish between what we might call advanced aging and accelerated aging.

So in typical aging, you might have a period of sort of stability across adulthood and then some kind of linear or perhaps, like this, nonlinear decrease as you get closer and closer to death, whatever it might be. Advanced aging would be that there would be an offset in every age. So at every age, the biomarker, for example, might look worse in people with schizophrenia than it does in typical aging, but the trajectories are basically the same. Still, when you have this kind of pattern, what you can see is that people are going to reach some kind of theoretical level of functional impairment earlier in life than they would do in typical aging.

Now, if you have accelerated aging, you might also have some advancement, but in addition you would see a faster rate of decline over time. So there you would get an even quicker advancement to functional impairment, but there would also be a more precipitous decline. This type of pattern of accelerated aging is something that's difficult to see unless you have longitudinal data.

You can look to see if there is a different relationship of a marker to age in people with schizophrenia compared to those without, but generally you need to have within-subject longitudinal data to really observe that kind of more precipitous decline that we would call accelerated aging.

So I'm going to be talking to you today about the systematic biomarkers of aging in schizophrenia, and this is motivated by looking at the types of things that happen when we get older. So the types of things that could be are things like genomic instability, epigenetic alterations, mitochondrial dysfunction, which we heard about yesterday, cellular and immune senescence, and telomere attrition, all of things contribute on a molecular level to aging and can be oftentimes seen in peripheral tissues like blood or stool.

One of the first things I want to talk to you about is inflammation. Inflammation is the body's natural response to insult or injury, and it's an important thing to have as part of the healing process. Although, what we know is as we get older, the proinflammatory response tends to increase and be more chronic as opposed to being a response to insult or injury, and some have called this the process of inflammaging.

So there are lots of contributors to inflammaging, and inflammaging has been seen in many different age-related conditions. We were interested to see whether we would see inflammaging in people with schizophrenia compared to people who do not have the disorder.

The study that I'm going to be presenting lots of results from is actually a longitudinal study of people with and without schizophrenia who range in age at the start of the study, and I'm going to be presenting mostly baseline data from this study. The average age of the sample is about 48 years. So it includes both younger and older adults with schizophrenia.

In this first study, we looked at cytokines, which are, again, a blood-based biomarker of inflammation, and we found elevations in two proinflammatory cytokines, tumor necrosis factor alpha and interleukin-6. Interestingly, when we looked at age relationships, we did not see a differential pattern of change across age groups in schizophrenia compared to people without schizophrenia.

We also looked at what inflammation was related to in terms of clinical variables and we found that it was related to more severe depression in both groups, and in the schizophrenia group to worse mental and physical well-being.

Another pro-inflammatory marker in the blood is high sensitivity C-reactive protein. We found elevated high sensitivity CRP in schizophrenia compared to healthy individuals. The levels were particularly elevated in women compared to men in the schizophrenia group, as you can see in this slide. The levels of HS-CRP were associated with metabolic risk, physical comorbidity, and with negative symptoms, but again, not actually related to age in either group.

Some other markers that we've examined include chemokines which are part of the innate immune system and are immune regulators. We found that MDC and eotaxin-1 together strongly distinguished between people with schizophrenia and people without and that they were related to the level of negative symptoms and also duration of illness within the patient group.

Dr. Lee looked at adiponectin which is a metabolic hormone and found that it was reduced in people with schizophrenia compared to healthy controls, and low adiponectin levels are generally associated with poorer health.

Interestingly, when we looked at age, for this marker, we found the lowest levels in younger people with schizophrenia, which made us wonder whether there was some sort of survivor bias such that those who were in our older sample were the ones that perhaps had the most favorable levels of adiponectin and that's why they had survived to be in our late life sample.

Another marker that we looked at was leukocyte telomere length which is a measure, as most of you probably know, of cellular damage. We did find that this was related to age in both people with schizophrenia and without schizophrenia, but not differentially related to age, and there was particularly a relationship to age in men within that combined sample. We also saw sex by diagnosis interaction in that women with schizophrenia had shorter leukocyte telomere length than women without schizophrenia, but there were no differences in length among the men.

So we had contributed a lot to the literature on systemic biomarkers of accelerated aging in schizophrenia, and we also wanted to see what the other literature said and what was out there in general. So led by Dr. Tanya Nguyen, we did a critical review of the literature, and we identified 42 articles on system biomarkers of aging, and 75 percent of these reported abnormal biomarker levels in people with schizophrenia compared to age batch controls.

However, only 29 percent showed that there was differential age-related decline that is showing. That is, showing that in cross-sectional studies, age was more strongly related to the marker in people with schizophrenia compared to people without. The studies that did find this pattern often were focused on either receptor or synaptic function or gene expression.

Finally, consistent with some of the individual results I just showed you, markers of disease severity and longer duration of illness were correlated with abnormal biomarker levels in patients with schizophrenia.

So the fact that we saw most promising findings in terms of differential age relationships in some of the more genetic and molecular markers that were in the literature, we turned to this as well by studying, in this case, proteomics. This was led by Dr. Campeau and Vivian Hook's lab, again using data from the same study that I told you about. This was looking at using an unbiased exploratory analysis of protein levels from plasma. What was identified were genes and proteins that were focused on inflammation and the metabolic system, and those were dysregulated in schizophrenia.

Also, when we looked at age effects, what we saw was that the schizophrenia proteome resembled that of older healthy controls, but younger healthy controls cluster all by themselves. This was true no matter the age of the person with schizophrenia. In fact, there was a cluster of proteins that showed this differential age effect pattern shown here where the youngest of the patients with schizophrenia actually had the highest relative abundance.

These proteins were also more abundant in the middle-aged and older groups with schizophrenia, and these were proteins that showed an increase with age in the healthy controls but not to such a same level. These proteins were characterized primarily as ones involved in the complement cascade, which is also part of the immune system, and regulation of that complement cascade, as well as the innate immune system which is consistent with our findings with chemokines and cytokines.

We were also interested in the role of the gut microbiome because, as you may be aware, the microbes in our gut play a role in inflammation and the production of short chain fatty acids. People with schizophrenia are known to have a more porous barrier between the gut and the rest of the body so that this leaky gut contributes to the output of the microbes and their actions being more likely to cross into the blood.

And so we wanted to look at the gut microbiome. This is work led by Dr. Tanya Nguyen who found that there was altered gut microbial beta diversity in schizophrenia compared to people without schizophrenia. In a very careful analysis, they identified lachnospiraceae as one of the microbes that had the highest increased abundance in schizophrenia. This has been associated with the production of butyrate and actually anti-inflammatory short chain fatty acids.

Several functional pathways were also found to be altered. These were ones associated with inflammation and cardiovascular disease.

So you can see that the results from the molecular, genetic, and gut microbiome studies are very consistent with what we've seen with blood-based biomarkers in terms of the pathways being related to inflammation and cardiovascular risk.

More recently, Dr. Ellen Lee has been looking at what kinds of modifiable risk factors, such as sleep, might be related to inflammation. In work that she presented briefly yesterday, she found that self-reported sleep quality, but not self-reported duration of sleep, was related to inflammation in schizophrenia, including finding elevations of both high sensitivity CRP and IL-6 in people who slept poorly.

Interestingly, we also found the same pattern in bipolar disorder and in people without any major mental illness using objective actigraphy measures of sleep variability. So this may not be a finding that's specific to schizophrenia. However, sleep is still a modifiable factor that could reduce inflammation in a group like schizophrenia where inflammation seems to be prominent and may even be there at levels that are close to causing morbidity, mortality, and functional impairment.

This is my second to last slide. There are still some gaps to our knowledge, however. As I mentioned before, the best evidence for accelerated aging comes from long-term longitudinal studies where we can measure within-subject trajectories of systemic biomarker levels, and we really don't have very many of those in the literature right now.

We also want to understand individual differences in trajectories. There is clearly heterogeneity across the older adult lifespan and trying to understand that heterogeneity will help us to figure out how we can change the trajectories.

Another missing feature is really we don't understand whether these systemic biomarkers are indicating something that's going to happen in the future or if they are a sign of something that's happened in the past. So whether they're leading or trailing indicators of decline in body and brain health is not clear.

We need to discover effective ways to modify these trajectories through, for example, altering sleep or altering diet in order to change the gut microbiome composition in order to prevent or slow declines in those people with schizophrenia who are the most vulnerable.

So again, I'd like to thank all of the collaborators, the participants in these studies, and the generous funding from NIMH that's made this work possible, and thank you for your attention.

DILIP JESTE: Thank you, Lisa. When I started this session, I was so excited to listen to the speakers that I forgot to introduce the speakers in the beginning. So here they are. You already heard from Lisa Eyler from UCSD on inflammaging.

The next speaker is going to be Dr. Dimitrios Kapogiannis from National Institute on Aging, and he will be speaking on exosomes, insulin, and ADRD mechanisms.

The next speaker will be Dr. Theo van Erp from University of California Irvine. He'll be talking on ENIGMA consortium, and last but not least, Dr. Hilleke Hulshoff Pol from Utrecht University in the Netherlands, who will speak on brain gap/age.


DIMITRIOS KAPOGIANNIS: Thank you so much for this kind introduction, and I am grateful for this opportunity to present in this panel, and I would like to thank the organizers.

I have done most of my work in Alzheimer's disease and neurodegenerative diseases. So I will share some insights in how I hope to convince you that some of them may be translatable to schizophrenia research. So my research has focused on extracellular vesicles. A few words about what are extracellular vesicles, and what is their pathogenesis.

When we are referring to extracellular vesicles, we're referring to a whole bunch of different vesicular structures that can be found in biofluids. The two main types are exosomes and microvesicles. Exosomes are the products of exocytosis from this structure, this intracellular structure called the endosome, which creates vesicles inside, we call them intralumenal vesicles, and the whole thing becomes a multivesicular body, and then some of these vesicles are sorted out in the surface.

Microvesicles refer to vesicles that are produced through direct budding of the plasma membrane. In reality, both types of vesicles can be found in biological fluids and form a continuum. Because of their dual biogenesis, these vesicles have a common structure and effectively can sample almost every part of the membrane, transmembrane proteins, and the cytosolic and intracellular signaling mediators. So they can be a great source of biomarkers.

My lab, we have been using neuronal and astrocytic brain-derived EVs as a source of liquid biopsy for neurodegenerative diseases and potentially also for psychiatric diseases. So brain cells -- astrocytes, neurons, microglia, endothelial cells -- like all cells for that matter, they release these various types of extracellular vesicles in the extracellular environment in the interstitial fluid. Some of these vesicles in a non-stochastic way have been shown through transcytosis and other mechanisms to find actually their way in the circulation, and there we can harvest them in peripheral blood using a two-part protocol where we first isolate total particles, total vesicles, through precipitation and then targeting specific molecules in their surface to give us subpopulations of vesicles enriched for a specific cellular origin.

In our lab, we've been using L1CAM as a neuronal marker to immunoprecipitate EVs of neuronal origin, and GLAST to immunoprecipitate EVs of predominantly astrocytic origin.

If you wonder how these vesicles look here, different types of electron microscopy images, this is a negative stain. You can see a variety of sizes and structures. These are typical EVs. CryoEM is actually a technique that allows to visualize these particles with great precision and to uncover their ultrastructural details, and you can see here in several examples that we have vesicles and the size distribution of smaller microvesicles and exosomes, and some of them are more exotic-looking. They have a vesicle within a vesicle double-membrane vesicle type of particles.

Recently, our approach to immunoprecipitate is based on L1CAM has come under some scrutiny because some investigators have questioned whether L1CAM is an effective molecule for pulling down neuronal EVs. So we have been generating evidence which we will publish soon demonstrating for instance with immune EM that immunoprecipitated vesicles as well as total vesicles can be shown to harbor L1CAM on their surface. Here we see some examples of L1CAM tagged with gold microparticles on the surface of vesicular structures. These are all isolated from peripheral blood.

Here with high resolution confocal microscopy, we can see co-localization of L1CAM with bona fide EV molecules such as Alix and the neuronal synaptic molecules such as VAMP, we can see that these neuronal and EV proteins co-localize in a very consistent way with L1CAM, pretty much demonstrating that neuronal EVs that express L1CAM exist, and giving proof of concept for our entire approach.

But over the years that we have, despite these doubts and the need to go back even to fill in some holes in the basic methodology, we've been using these vesicles to generate a wealth of biomarkers for Alzheimer's disease which we hope inform an approach of precision medicine, looking at different pathways. So we've been able to develop biomarkers reflecting amyloidosis such as A beta 42 and other A betas as cargo of these EVs, tauopathy, total tau and phospho-tau, neurodegenerative-related molecules, multiple synaptic proteins that can speak to synaptic degeneration in AD and other diseases, a wealth of neuroinflammatory mediators.

When I refer to neuroinflammation, I predominantly refer to not cytokines, but intracellular mediators of cytokine signaling. This is what EVs are actually best in doing, as well as multiple metabolic mediators in the insulin pathway, but also recently mitochondrial components.

So the very first study of neuronal EVs actually focused on deriving diagnostic biomarkers for Alzheimer's at the MCI early dementia stage, and as it was naturally we focused on amyloid and tau protein. What we were able to show with that if we compared in a case control study of AD versus control participants, AD individuals had much higher level of phospho-tau and A beta 42 than controls.

The second pathway actually we looked at was again to derive diagnostic biomarkers for AD at the early clinical stage, was the insulin signaling pathway, and our wish was to derive biomarkers for insulin resistance. When we refer to insulin resistance, we refer to decreased signal propagation downstream of an insulin receptor. The first node in the insulin signaling pathway is IRS-1, insulin receptor substrate 1, and there are a number of phosphorylations that modify the ability of this molecule to function and propagate the signal downstream. Tyrosine phosphorylations in general are considered as enabling, enhancing insulin signaling. Serine phosphorylations are considered generally as representing a blockade to insulin signaling, and marking resistance.

In the Alzheimer's disease, it had been shown that AD patients compared to controls and MCI show a dramatic accumulation of pSer-IRS-1. We were able to use our neuronal EVs as a type of liquid biopsy biomarker to demonstrate that, whereas total IRS-1 is similar between AD patients and controls, serine phosphorylated IRS-1 showed a similar dramatic increase as do path sections.

Furthermore, we're able to show that these IRS-1 biomarkers relate with brain volume on MRI. So here in the first study published in 2017, we showed that the good IRS-1 which marks increased insulin signaling in neurons is associated with more preserved volume. And then if you notice the pattern of this VBM association, one sees that it is predominantly on the lateral temporal lobe, so that triggered our interest, and this is a finding we're actually able to represent, to reproduce in an entirely different cohort of aged individuals and MCI individuals where the temporal lobe volume was again positively correlated with neuronal p-tyrosine and IRS-1.

On the contrary, neuronal tyrosine IRS-1 was negatively associated with the volume of white matter hyperintensities, suggesting that neuronal insulin signaling is good for preserving gray matter and for preserving neuronal microvasculature and preventing the appearance of these white matter hyperintensities.

Then we took baby steps venturing into the psychiatry work and did a collaborative study with Johann Steiner and colleagues from Germany. We investigated whether our insulin signaling biomarkers may have some prognostic and diagnostic ability in schizophrenia, and in an early case control study, we saw that downstream mediators in the insulin-signaling pathway showed a pretty small but consistent decrease in phosphoprotein expression when we're looking at downstream mediators such as Akt GSK3Beta, mTOR, and S6K, which is downstream of mTOR, we see a consistent decrease in schizophrenia compared to controls.

The insulin signaling biomarkers took a life of their own and the psychiatry field, and this is a study which was performed without our involvement by Dr. Natalie Rasgon and Karl Nasca where they looked at the depression and depressed individuals, male and female, and found that individuals with depression, female individuals with depression, had higher levels of base insulin resistance marker, even though the same was not seen in men.

Most recently, we've been very interested in the mitochondrial abnormalities in AD and psychiatric diseases, and I was very pleased to attend several pioneering and amazing talks yesterday on mitochondrial abnormalities in schizophrenia. In a recent study in Alzheimer's disease which have shown, a, that electron transport complexes I through V, all the way from complex I, complex III, complex IV, and ATP synthase otherwise known as complex V, are all decreased in AD patients compared to controls.

Remarkably, we were able actually to quantify the catalytic activity of these molecules here shown for ATP synthase and complex IV activity and show that these molecules are not only present in lower quantities in EVs, but their catalytic activity is also compromised and we see decrease in ATP synthase activity and complex IV activity in AD compared to controls.

In collaboration with my colleague and mentor in this field, Ed Goetzl, and other outstanding colleagues, we've been looking at mitochondrial proteins in depression. Here is a study where depressed individuals were broken down by treatment responders and non-responders, and we were able to assess their levels of these various mitochondrial proteins, both baseline and after response, and what Ed and us with colleagues have shown that many of these mitochondrial proteins had actually shown response to treatment and normalized after successful antidepressant treatment.

Recently Ed took this further and looked, without my involvement in that this time actually, at these same mitochondrial electron transfer proteins that I referred to earlier and investigated them in both neuronal and astrocytic EVs in control versus first psychosis individuals and has shown that their levels are decreased in first psychosis individuals, both in neuronal and astrocytic EVs. These are a number of proteins in the electron transport chain, SOD1, which neutralizes reactive oxygen species as well as some very interesting mitochondrial small peptides that may have neuroprotective function, humanin and MOTS-c.

And as a sign of very convergent evidence and validating these findings, when I was preparing with this for this talk, I saw that actually this was just already reproduced. Here was a study looking at both an animal model of psychosis as well as patients, and it was shown that this component of ATP synthase, COX6A2, was shown to be decreased in EVs of these animals and these researchers actually referenced our previous studies and took note of L1CAM as a means of localizing these neuronal EVs and demonstrating that they exist, and they showed that in early psychosis there is a decrease in this mitochondrial alongside other proteins, and also found that there is a correlation between their levels of this mitochondrial peptide with ERP auditory potentials across patients.

So I will just like to finish with just another dimension of all this. We've talked about the ability of EV biomarkers to distinguish between patients and controls, but I actually, I think that perhaps their greatest significance may be in demonstrating target engagement in clinical trials, and may represent messages in a bottle that may allow us to see sort of response to experimental treatments. The first study in that regard, we looked at the response of the insulin signaling cascade to an antidiabetic medication, exenatide, in a trial in Parkinson's disease patients. This was a positive trial, by the way, showing that exenatide had a substantial benefit in PD.

There we were able to show that the exenatide treated compared to placebo treated individuals showed an increase in the p-tyrosine IRS-1, as would be expected by enhanced insulin signaling compared to placebo treated individuals, that was maintained even at washout, and this response was a concerted response, not an individual random molecule goes up type of response, but it was a response that was propagated across the entire cascade showing elevation in AKT and mTOR and other molecules along the cascade.

And most recently, we were able to show that this promise may actually hold also for schizophrenia. We leveraged samples from a study published by Roger McIntyre and colleagues in JAMA Psychiatry where they administered infliximab, an anti-inflammatory anti-TNF-alpha antibody, to patients with bipolar disorder, with or without physical abuse, and they showed that there is a responder subgroup, those bipolar individuals with physical -- who had history of physical abuse, showed a significant response to infliximab.

And we were able to show that at the molecular level, leveraging their neuronal EVs, in fact if we look at the mechanism of action in infliximab goes through TNFR, TNF receptor, and these inflammatory signaling mediators of the NF-kappa-B pathway, and we were able to show that individuals with history of physical abuse showed a response of these TNFR and NF-kappa-B mediators, which their non-responder, clinically non-responder subgroup, did not show.

This correspondence between clinical response and biomarker response I think is what is important for clinical trials in the future, and what EVs can make a difference in.

So conclusion. Neuronal and astrocytic EVs contain biomarkers that may reflect the different aspects of disease pathogenesis, which can result in multidimensional patient characterization in neurology and psychiatry. That may suggest overlapping mechanisms between neurodegenerative and psychiatric diseases, and so we need to continue learning I think from each other and reading each other's literature.

Mitochondrial abnormalities appear in many diseases, and I'm questioning here whether they're causal or a nonspecific marker of brain vulnerability.

There is continuous need for methodological innovation in the EV field and further advance these EV isolation techniques. I envision a future where we're going to be able to isolate dopaminergic, serotonergic neuronal EVs and that that will enhance our capabilities to discover salient biomarkers.

We need to tie in lipids, RNAs, DNAs, and other functional assays, and we can use EV biomarkers to select subgroups for clinical trials and to demonstrate target engagement and predict treatment responses.

The ultimate vision is for precision medicine clinical trials, a future with precision medicine clinical trials, where the right intervention is applied to the right subgroup of patients at the right time in their evolution of the disease, and responses are evaluated using the right outcomes, not just generic scales or clinical responses.

With that, I would like to thank the hardworking members of my lab, my multiple collaborators from across the world, including some psychiatrists, which I have been trying to work and learn from, as well as Dr. Ed Goetzl, who has been a pioneer in these studies.

Thank you.

DILIP JESTE: The next speaker is Theo van Erp. Theo, would you like to start?

THEO VAN ERP: Good day, everyone. I want to thank the organizers for inviting me to present some of the findings from the ENIGMA schizophrenia working group at this workshop.

I have nothing to disclose. All of the work presented has been supported by the NIH and NIMH.

I'll start with a brief introduction to the Enhancing Neuroimaging Genetics through Meta-Analysis consortium, which most of you may be familiar with. ENIGMA was started in 2009 by Jason Stein and Paul Thompson to collectively analyze worldwide brain imaging and genetic data.

By the time of its first publication, we realized that about one third of the data contributions came from clinical disorder studies. Around the same time, there were numerous studies reporting on a crisis of replicable findings in psychology and neuroscience. Under the leadership of Paul Thompson, the consortium applied for funding by the National Institutes of Health's Big Data to Knowledge award in 2014. The initial project included several methods and nine disorder working groups, including this schizophrenia working group.

Since then, ENIGMA has grown to a consortium with more than 50 working groups including numerous methods, developmental, and clinical and nonclinical working groups that collectively pull together more than 80,000 brain imaging data sets including MRI and EEG from more than 30 brain disorders. The consortium has a federated organizational structure with groups chaired by scientists from different countries and has grown to include more than 1,400 members from 400 institutions across the globe.

Together with Dr. Jessica Turner from Georgia State University, I co-chair the ENIGMA schizophrenia working group. The initial goals of the ENIGMA disorder working groups, and this is also the schizophrenia working group, included the ranking of case-control or disease-related effect sizes of brain measures, the examination of factors that may contribute to moderating these effects, and comparing findings across disorders to answer the question of shared versus disease-specific brain abnormalities.

To get this effort off the ground, we started with the lowest hanging fruit which was the examination of deep brain structure subcortical volumes, because numerous samples had already analyzed their imaging data for the imaging genetics studies.

So in our first schizophrenia working group study, we reported on rank order case-control effect sizes for deep brain structures. In the meta-analysis, including more than 2,000 individuals with schizophrenia and 2,500 healthy controls, we found significantly lower hippocampal, amygdala, thalamus, accumbens, and intracranial volumes, and larger pallidum and putamen volumes and lateral ventricle volumes in individuals with schizophrenia compared to controls.

In a parallel study, Drs. Okada and Hashimoto and colleagues replicated the rank order effect sizes for deep brain structure volume abnormalities in schizophrenia in the Japanese COCORO consortium using similar analysis procedures. This was very exciting to both our consortia as we now each have a collaboration through which we can generate independent replications of clinical neuroscience findings in schizophrenia.

This study provided the first ranking of case-control effect sizes, but how about variables that moderate these effect sizes? In our initial analysis, these were examined using meta-regressions. We found that mean pallidum and putamen volume effect sizes, basically, were positively associated with duration of illness and also age. In addition, we found that the severity of hippocampal volume deficits in schizophrenia was negatively associated with the proportion of not medicated patients. These effects must be regarded with caution as meta-regressions are prone to statistical artifacts, but do suggest the ability to examine effects of factors that may moderate the observed brain effects.

The working group recently followed up on these findings with shape analysis of deep brain structures, which allow for more regional specificity of the observed volumetric group differences. Consistent with the observed volumetric findings, these analyses revealed more concave than convex shape differences in the hippocampus, amygdala, accumbens, and thalamus, more convex than concave differences in the putamen and pallidum, and both concave and convex shape differences in the caudate in individuals with schizophrenia compared with healthy volunteers, some of which were associated with severity of positive symptoms and chlorpromazine dose equivalents.

With regard to cortical abnormalities, we examined cortical thickness and surface area abnormalities in schizophrenia. On these three-dimensional cortical surface models, the more red compared to yellow colors reflect thinner cortex compared to controls, and the largest effects were observed for the fusiform gyrus and inferior, middle, and superior temporal lobe gyri, superior frontal gyrus, pars opercularis, and bilateral insula.

When statistically controlling for global cortical thickness, several frontal and temporal lobe case control differences remain significant, suggesting regional specificity of cortical thickness abnormalities in schizophrenia. As I mentioned, in the analysis statistically controlling for cortical thickness, we found that several regions show more than average lower thickness while lower regions show less than average lower cortical thickness, indicating that the pattern of abnormalities is regionally specific.

Like the subcortical volumes, these cortical thickness abnormalities can be ranked based on their severity, and because ENIGMA working groups aim to use similar methods across working groups, profiles can also be compared across disorders. In the review that's currently in press at Psychiatry and Clinical Neurosciences, we compared imaging methods across several disorders, and the pattern of cortical thickness deficiencies, for instance, shows the highest correlation between schizophrenia and bipolar disorder, which are known to have phenomenological and genetic overlap.

In terms of moderating factors on the effect sizes of cortical thickness, we observed large effects of medication treatment at the time of scan. Compared to unmedicated patients, the effect sizes for case control differences in cortical thickness were twice as high in patients treated with second generation antipsychotic medications and three times as high in patients treated with first generation, or both first and second generation antipsychotic treatments.

We did not observe significant group by gender interaction effects in cortical thickness. However, age of onset and duration of illness were associated with insula cortical thickness, and there was a significant group by age interaction on temporal pole thickness, with a steeper negative association in schizophrenia compared to the control group.

We also found significant associations with standardized medication dose, chlorpromazine equivalents, even when controlling for total negative symptom severity, as well as total negative and positive symptom severity. We are in the process of exploring these symptom associations in more detail.

With regard to cortical surface area, we found globally smaller cortical surface area in individuals with schizophrenia compared to healthy controls with effect sizes that are about half of those observed for cortical thickness. We did not find significant relationships between clinical factors and cortical surface area.

A comparison of regional white matter microstructure based on diffusion weighted imaging data between more than 1,900 individuals with schizophrenia and more than 2,300 healthy volunteers found significantly lower fractional anisotropy in 20 out of the 25 white matter tracts examined. However, when controlling for global fractional anisotropy, no region showed significantly lower FA, suggesting global widespread effects on white matter microstructure in schizophrenia.

With regard to the question of accelerated aging, the working group under the leadership of Mr. Constantinides and Drs. Walton and Dima recently examined the difference between brain age as estimated based on brain morphological measures and chronological age and found that brain age in individuals with schizophrenia is on average 3.64 years older than their chronological age. However, brain age was not significantly associated with any clinical characteristics such as age of onset, duration of illness, severity of symptoms, or treatment. For instance, medication type or chlorpromazine equivalents. This suggests that brain age in schizophrenia is not primarily driven by disease progression or treatment-related effects on brain structure.

As mentioned, one of the advantages of using similar methods is the ability to compare findings across disorders. The 22q11 deletion syndrome working group, for instance, was able to compare the cortical thickness patterns in individuals with 22q11 deletion syndrome and psychosis with those without psychosis to the schizophrenia and major depressive disorder working group effect sizes and showed that this pattern was more similar to that of idiopathic schizophrenia.

In addition, the ENIGMA clinical high risk group also examined correlations between regional cortical thickness effect sizes in clinical high risk converters and compared those from schizophrenia and the 22q11 deletion syndrome psychotic patients and found positive correlations with both. These findings suggest that the cortical thickness is replicable and disease-related biological abnormalities in schizophrenia.

Summary and conclusions. Schizophrenia is associated with replicable deep brain structure volume and shape abnormalities. Schizophrenia is associated with robust regionally specific cortical thickness deficiencies. Schizophrenia is associated with global cortical surface area deficiencies that are about half the size of the observed cortical thickness deficiencies. Schizophrenia is associated with global white matter microstructure deficiencies.

Medication type and dose are associated with lower cortical thickness but not surface area or fractional anisotropy. Insula cortical thickness is associated with age at onset and duration of illness. Temporal pole cortical thickness shows a stronger negative association with age in schizophrenia compared to controls, suggesting that if any differential aging effects are present at the structural level that they may be strongest in the temporal pole.

Predominantly frontal and temporal lobe cortical thickness are significantly negatively associated with total, negative, and positive symptom severity. The cortical thickness effect sizes in individuals at clinical high risk for psychosis, 22q11 deletion syndrome with psychosis and bipolar disorder, are correlated with those of idiopathic schizophrenia or psychosis, suggesting that the distribution of effect size is reflective of underlying biological abnormalities, and given the genetic overlap between schizophrenia and bipolar disorder, at least likely in part due to genetic factors.

Last but not least, I'd like to thank all of those who contribute their time and effort and expertise to ENIGMA overall and particularly to the schizophrenia working group. I'd like to thank our funding sources, the NIH and NIMH. It's really been a tremendous privilege and joy to work with so many excellent collaborators around the world. Thank you for your attention and I'll pass the torch to Dr. Hilleke Hulshoff Pol for the next presentation.

HILLEKE HULSHOFF POL: Thank you, Dr. van Erp. Thank you. Today I would like to talk about schizophrenia and brain aging and age/gap, but first I would like to thank the organizers for a wonderful and very impressive workshop as well as the providers; I am very grateful to the grants for the work I'm going to present as well as the participants in those studies as well as my coworkers, in particular Hugo Schnack, Jalmar Teeuw, Rachel Brouwer, Rene Kahn, for studies that have been conducted at University Medical Center Utrecht and Utrecht University, and you may appreciate immediately what you see on this slide, and that is people biking and walking. I would like you to keep that in mind because at the end of my presentation, I will get back to that, since it seems to be beneficial to brain change.

Not only during childhood and adolescence, but also in young and middle adulthood as well as in older adulthood, our brain changes. Our brain changes throughout life, and you can measure that in several ways. Obviously there's no need to explain why it's so important to know more about how the brain changes, both for aging and in particular for people with nonaffective psychosis, and you can do it several ways and we've seen beautiful already examples of cross-sectional studies looking at the changes with age.

The way we do this is by measuring with brain scans using MRI to induce two scans per individual which allows, if you will, trajectories, but it least it can show you change within individuals. That's measured as a delta change.

Now this for example is a meta-analysis of over 2,000 scans on change with age, longitudinally, in whole brain volume. What you see here on this line is between the age of 20 and 40, there's not much happening. It's quite stable. Whereas between 40 and 60 years of age, there's a subtle but steady decrease which accelerates somewhat after the age of 60, but please appreciate there's still 1,000 milliliter on the zero on the y-axis left. So it still changes by the age of 90, there's still at least 85 percent of brain tissue left. But there's a decrease.

It also differs per structure, the trajectories of the change. So for instance, cortical volume decreases already from childhood on, whereas white matter increases throughout at least until 40 years of age, and then slowly starts to decrease. So these trajectories are individual per structure.

What happens in schizophrenia? People diagnosed with schizophrenia are shown on the red line in compared to the controls who are healthy controls shown in black here. What you can appreciate between 20 and 40 is quite an accelerated decline in whole brain volume. Whereas if you put a straight line through it, that is. But if you then look at what happens between 40 and 60 and possibly beyond is that it is quite parallel. There not seem to be much of a difference between the way there is change in schizophrenia as compared to a normal aging. It seems to have the same pattern. But please, of course, as you will see, there is a gap between the two.

How does that look? There's also individual differences between individuals. On the right, you can see a healthy individual around the age of 40 years. In the middle there's a patient with schizophrenia around the age of 40, and the left as well, except that on the left there's a clear atrophy of brain tissue whereas in the middle there's a brain which cannot be distinguishable from normal. So these differences can also differ. They have a relationship with some kind of functioning, but there's also individual differences.

Now does this reflect accelerated aging of the brain, or this caused by a fundamentally different process? We set out to do this using machine learning in which we measured brain age/gap and brain age/acceleration. So how was this done? Based on the gray matter voxels, a model of the brain was made based on the healthy -- the MRI scans of the healthy individuals, and this model then, each person was then or the MRI scans of each individuals were then set against this model to determine brain age. So for instance, if somebody was 60 years of age and their brain age was 55, you had a brain age gap of -5 and your brain looks somewhat younger, 5 years younger, than your true age.

In contrast, if you had a brain age of 65 years, then it would be different, and if your brain age would be 60 whereas you had a true age of 60, it would be a zero gap. In other words, there is an indication which is quite static.

But then the acceleration, the multiple scans we use, there was a second scan done. So on top of what you have in this gap, for instance, if you look on the right, there's this green line with normal aging trajectory on the x-axis, the true age and the brain age on the y-axis, and then in orange you can see the increased aging, but no acceleration. Early it was used I think advanced aging as well. But then the accelerated age is on top of that a faster aging.

Now what was found in this study in approximately over 700 individuals that brain age gap was in schizophrenia plus 3.36 years greater than chronological age, quite comparable to what Dr. van Erp mentioned earlier in this huge ENIGMA study, and brain age acceleration on the right was 2.5 years early after illness onset. Please note that 2.5 times faster than normal, whereas within 5 years this was normalized towards one year per year, which you would expect with true aging.

So it seemed to be largely around early phases of the disorder and less so later on. Please also appreciate again the individual differences which we have to keep in mind that some patients showed a much younger brain compared to the true age, whereas also some controls showed a much older brain compared to their true age. So these were the average findings.

Now we were wondering what could cause or at least be associated with those changes. So we asked the questions do individuals who carry high schizophrenia polygenic risk also present with a faster brain age acceleration? So polygenic risk, as you know, is a combination of weighted genetic load, which is associated with the risk to have a diagnosis of schizophrenia. We wanted to know if this was related to these changes. Included were over 400 on the y-axis subjects of which we knew on the x-axis is the chronological age or true age, with illness onset information, two MRI scans, and genotyping in blood samples. We also did epigenetic blood samples, but I will focus on the genotyping here.

What did we find in this second study? We found on the zero axis you can see on the left what you would expect from the model, right? So there's no gap in the healthy individuals, and there is an increased MRI brain age gap in patients with schizophrenia, which are on this violin plot, the darker grayish colors.

There's also acceleration. There's acceleration in healthy individuals. There's already some speeding up compared to the gap that they started from after those years, and that is somewhat more pronounced in patients with schizophrenia.

And then in this group, there was a polygenic risk in blood identified which was not increased in healthy individuals and increased in the individuals with a diagnosis of schizophrenia.

Was there an association between the two or three? Yes, there was, a very subtle one, but it was significant in the sense that MRI brain age/gap was associated with polygenic risk for schizophrenia, as well as MRI brain age acceleration was positively associated with polygenic risk for schizophrenia. In other words, an increased risk for the disorder was associated with an increased, with higher brain age and more acceleration.

However, if we controlled for diagnosis, then this effect no longer was significant. At least there were only 400, which is kind of small for these kinds of study. The reason I think this may still hold up is that in a recent study we concluded in the ENIGMA working group on plasticity, we completed a study in 15,000 individuals with two MRI scans and did genotyping to find genes for brain change. There there was pleiotropy between genes for schizophrenia, with genes for brain change. So I think there is an overlap still and that it's very relevant to look for mechanisms and also have a hold on to assist in individual treatment in the future hopefully. So keep that in mind.

Now is there anything in the environment or in that can be done for these changes? This goes back to the first slide I showed. We asked the question does exercise of an overland physical skill improve brain connectivity? So here we looked not at the gray matter but at the white matter, so the connections between the distant anatomical brain areas using DTI, and we looked at fractional anisotropy.

Again, this was a longitudinal study, and here we asked both patients with schizophrenia, approximately 40, and healthy individuals, approximately 40, either to have life as usual or the other half to do an exercise trial consisting of biking twice weekly for six months. In fact, it was three-quarters of an hour biking on a stationary bike and 15 minutes of lifting weights.

We did an MRI scan before and after this intervention. Now if you look on the right, top right, there's a non-exercise group consisting of controls in green and the patients in orange, and you see that at that age, around the age of 40 years, there's already a subtle decrease in fractional anisotropy as a measure for integrity of the white matter network.

With exercise, in fact, this tend to go up, both in patients with schizophrenia as well as in healthy individuals. So we then advanced this together and looked where this actually was situated, which fibers were involved, we find that in red showing the increase in fractional anisotropy and in blue the decrease in fractional anisotropy, with on the top the exercise group and in the bottom the non-exercise group, that in the exercise group there is an increase, and this is particularly in the corticospinal tract. This corticospinal tract connects the brain with the rest of the body, in particular the arms and legs, which are actually obviously busy during those exercises. So I think this is a very hopeful point.

With that, I would like to conclude that human brain changes throughout life, in part under the influence of genes. In schizophrenia, brain age was greater than chronological age. The acceleration of aging is most pronounced in the early phases of the illness, in part due to increased genetic risk, and exercise of an overlearned physical skill improves brain connectivity.

With that, I would like to thank you for your attention and give the floor back to Dr. Jeste.

DILIP JESTE: Thank you very much. This has been a great session full of outstanding presentations. There are lots of questions. So what I would like to do is actually go through questions that are specific for each presenter. There are some questions that are general, and we may want to save them for the panel discussion.

So let me start with a question for Lisa. You used the word advanced aging. Is that different from accelerated aging?

LISA EYLER: Yes. I think Dr. Hulshoff Pol's presentation really illustrated that quite well. I think that there can be advanced aging where people the same age have an older-looking biomarker profile than people without schizophrenia, and it may be the result of an acceleration earlier in life. So what she showed was that closer to the onset of illness, there may be an accelerated pattern, but then you kind of get stuck at that point of worse brain or worse biomarkers, and then from then on out, there may be no difference in the profile of aging, no acceleration there.

Or there could be, in addition, more of a precipitous decline in people in their biomarkers, and I think we just unfortunately don't have a lot of longitudinal biomarker studies to help us understand what's going on just yet, and in the brain imaging work, it looks like that there is more stability and not an accelerated rate of decline after a certain age.

DILIP JESTE: Thank you. Dimitrios, a question for you. You talked about Akt and GSK3Beta as important molecules related to insulin. The question is that they are also related to cancer and antipsychotic use. So what are your thoughts about that?

DIMITRIOS KAPOGIANNIS: First, to clarify, whereas I and others when we present talks and focus on one cascade, we show these neat diagrams that the signal goes from the receptor to Akt and then downstream. It seems so neat and clean. In reality, there is a convergence of multiple pathways from multiple trophic factors and these cascades, and especially both Akt and GSK3Beta are under the collective influence of multiple signaling cascades, specifically IGF1 cascade greatly overlaps with that of insulin, and that may have relevance to some cancers, as well as many other neurotrophins and whatnot.

So that's why I think that these molecules have some values as diagnostic biomarkers, perhaps if there is a clear difference, but then even if there is a clear difference in diagnostic capacity, we don't really know what is upstream and what is downstream, unless you just look at every signaling pathway, which you cannot do.

I think they're actually more useful in treatment trials where you know the perturbation, where you know the mechanism of action of the agent that you introduce in the system, and then you're able to say, okay, everything else, like everything else staying the same, if I introduce a perturbation in the insulin signaling cascade or the anti-inflammatory cascade or what is the outcome for these markers?

DILIP JESTE: Thank you. Theo, you described the ENIGMA consortium. Actually I like that word, because I remember reading about schizophrenia as an enigma wrapped in a mystery inside -- a riddle wrapped in a mystery inside an enigma. So your consortium will hopefully solve the enigma of schizophrenia.

You presented some very interesting findings. The question is sort of what do you see as still the gaps that are there in the literature overall and in the ENIGMA work?

THEO VAN ERP: So I think there are several gaps still. One of them is that I think in terms of studies that are available is the number of longitudinal studies, which I think Hilleke Hulshoff is really addressing across multiple different disorders and samples, but I think even in schizophrenia, I think there are still relatively few.

Also, the age range I think is limited. Our average age for the samples that we have are around 30, with maybe the mean age between 20 and 40, and there are a lot fewer in elderly. So I think those two, and then of course there are lots of additional types of analyses that can be done with this type of data, including like network-based analyses and resting state analyses to see if those are more sensitive to change over time.

DILIP JESTE: Thank you. Hilleke, you talked about the polygenic risk factor, and the question is which genes do you think are specifically related to accelerated aging of the brain in schizophrenia?

HILLEKE HULSHOFF POL: This is an excellent point, of course, but that's still also an enigma, unfortunately. But now since we have these identified several genes in this large ENIGMA plasticity study in over 15,000 individuals, which also include some patients with schizophrenia, I think we can now go look for the overlap and see if we can get closer to finding which genes this could be. So that is work for the future, yeah.

DILIP JESTE: Thank you. There are also some other questions, and actually I will welcome the speakers to choose the question that they want to answer in the Q&A box. So Lisa, you want to take up any question that you want to address?

LISA EYLER: Sure. Actually there was one that came to me from one of the other panelists that was asking the question about kind of emphasizing heterogeneity in these trajectories of biomarkers and I think that we are at a point right now where we can start to move beyond just saying on average people with schizophrenia have this elevation and try to understand are there subgroups or clusters of people for whom, say, inflammation is a big part of their pattern and predicts something important about their prognosis, and if so, are there particular treatments that might be helpful for them?

So it was pointed out that there are some studies where the treatment doesn't seem to work overall, but might have worked if we targeted them to particular people who need it, and we could take a clue from some of the studies of, say, for example, infliximab in major depression where overall there was no positive effect on symptoms in the group. However, a subgroup of people who had high levels of inflammation at the beginning of the study did seem to show a positive effect. So I think we need to understand what are the different subgroups of people in terms of systemic biomarkers and what does that then imply for how we would treat those individuals?

DILIP JESTE: Thank you. Dimitrios, I have a question for you. Last year, Ellen Lee was the first author of a paper that we published in collaboration with Robert Rissman from UCSD. It looked at neuron-derived and astrocyte-derived exosomal amyloid beta tau and phosphorylated tau. It seemed to suggest that astrocyte-derived exosomes and the biomarkers involved were more relevant to schizophrenia than neuron-derived ones. Do you have any thoughts about that?

DIMITRIOS KAPOGIANNIS: Well, I am not so knowledgeable about the implication of astrocytes in schizophrenia. I would imagine that they are implicated deeply just because of everything the neurons do, it's because the astrocytes support them. It wouldn't surprise me. So I don't know the details of that paper. I'm sorry.

DILIP JESTE: That's okay. We'll send it to you. Thank you.

Theo, do you have any question you want to pick up from the questions asked?

THEO VAN ERP: I am actually not able to access the chat box for some reason. When I click on it, it doesn't show.

DILIP JESTE: Okay. Maybe Hilleke, in the meantime, do you want to take up a question?

HILLEKE HULSHOFF POL: Yes, I think that one that also actually is asked to Theo is about the biological aging. Older people in schizophrenia, if there's any other explanation for that. I think that Theo actually made a very important point stating that what we're often studying at least using imaging is individuals who have been ill for many years before, and what I've learned very much the past two days is that it is so important also to see if individuals, patients who become ill much later, and to see how their trajectories would go and if that would have different mechanisms for instance. So I think that's very important to pick up, because that might other different biological processes.

DILIP JESTE: Thank you. Theo, one general question but it also applies to ENIGMA study; Carl Cohen asked about the socioeconomic status, that when we compared people with schizophrenia and normal controls, we think that the main difference is because of schizophrenia and maybe antipsychotics. However, there's a big difference in the socioeconomic status and the social determinants of health clearly have an impact on disease research, as biomarkers. Are you taking a look at them in the ENIGMA study?

THEO VAN ERP: Yes, there is a group currently taking a look specifically at socioeconomic status. They're still in the process of pulling together data. So but it will be looked at, yes.

LISA EYLER: Dilip, I just wanted if I could just to follow up on that, because there was also a question in the Q&A about who gets left out of these kind of studies, whether it's brain imaging studies or biomarker studies, and it is true that our results or generally the samples that we have are generally not representative of the whole population. They tend to be WEIRD; you have probably heard this acronym of western, educated, industrialized, rich and democratic. The samples tend to be highly skewed in that way, and one of the questioners also mentioned that we often don't have homeless people in our sample. We don't have folks who have anosognosia, et cetera.

So there are a lot of ways in which our sample is enriched for people who are functioning the highest. So I think that's why it's exciting when we have studies like the one that we saw yesterday from rural China where we go out to people and try to bring our techniques to where people are, and I also think some of the techniques that we're using using mobile phones and actigraphy watches, where we are looking at people where they're at, can be very helpful to improve the diversity of the samples and the generalizability of our results.

DILIP JESTE: Thank you. So maybe we will switch to the last session within our group. So this is panel discussion, and welcome the co-chairs, Ellen and Sophia. One general comment I want to make and get some thoughts about that and then we'll open it up for this general questions.

We have been talking about brain aging, and yesterday there was a lot of discussion about dementia. So there is no question that people with schizophrenia have progressive cognitive impairment, more than we see in healthy subjects. Schizophrenia is clearly a brain disease. The question is it only a brain disease? Is it not a disease of the whole body? Because if you look at the causes of -- mortality in schizophrenia, we all know is high. People with schizophrenia die 15 to 20 years younger than the controls. Again, Ellen was the first author of a paper that was published a few years ago that showed that the mortality gap between schizophrenia and the general population has increased in the last 40 years, and it has not increased because people with schizophrenia are dying faster. It is increased because the population in general is living longer. But there has not affected people with schizophrenia. Their death rate continues to be the same.

And what do they die from? Typically, the causes of death in schizophrenia are same as the causes of death in the general population. Heart disease. Metabolic diseases. Cancer. The only difference is that these diseases are currently younger age. The chances of having an MI at 75, a person with schizophrenia develops the MI at 50 or 55.

And at the time, their heart and liver and kidney, they really look diseased. On the other hand, if you look at the brain in a person with schizophrenia, I don't think it shows changes that we see in Alzheimer's or Lewy body dementia or frontotemporal dementia. There is no obvious atrophy of a large set in an individual person. There is a group difference, of course, schizophrenia versus controls, and we see difference in ventricular size, et cetera. But an individual person with schizophrenia doesn't have high risk of developing that kind of brain atrophy.

What do people think about that?

SOPHIA FRANGOU: I will have a go, Dilip. The first part of your question said is schizophrenia not a disease of the whole body, and I would like to turn this around a little bit to say is there any disease that is not a whole body disease? If you start with diabetes, you end up in the brain. If you start with cancer, you end up in some ways in the brain. If you start with cardiovascular disease, metabolic disease, you end up in the brain, and so on.

So I think if anything, we can congratulate ourselves as psychiatrists that we have introduced this notion of whole body and biophysiological kind of interactions much earlier than our colleagues in medicine who are just beginning to realize that happier people with MS also have better outcomes and social support and activity, health, everything even in neurodegenerative disorders. So I think the body is an integrated system, and of course there is no cutoff at the neck.

That increases the complexity that we are faced with really, because signals from the body affect the brain and so on, and I think that it's very difficult to see in such a very complex and sort of chaotic system, not that it doesn't have any order, but it's difficult. The classic definition of chaos is it's difficult to predict the outcome. It's really difficult to see how best to intervene, and that to my mind sort of speaks to increasing really sample size when you have such complex interactions. You cannot do it with 80 or 100 people. So that's kind of an important message. And of course, longitudinal sort of follow-ups and so on.

Now it is absolutely true that the brain of patients with schizophrenia does not show the dramatic atrophy that you see, for example, in Alzheimer's, although I have to qualify that that we have very few imaging studies of patients age 70. So it could be that we're missing something, but in principle I'm happy to accept that it won't be that sort of dramatic.

And I think that it possibly -- and I think Roman has suggested some hypotheses as to why there might be dementia -- that I wouldn't call it dementia, although sort of classically if you use sort of the rating instruments for cognitive impairments, it's dementia. But I would call it like hugely unsuccessful aging rather that sort of a specific dementing thing. But that is just my take. So I'll stop here and see what other people think.

LISA EYLER: I was just going to say that, again, to emphasize that in some ways it doesn't really matter whether going back to the question of is it advanced versus accelerated or is it -- are the things happening in the brain going at the same rate as things elsewhere? Really what matters is when do people with schizophrenia end up at a point where they're functionally disabled and what can we do? What are the modifiable risk factors that we can do to help them to not reach that period of that point of highly unsuccessful aging, as you say, Sophia, as early as they are right now?

So how can we extend the quality of life as well as the quantity of life in people with schizophrenia, whether that's by altering some of the fundamental processes that contribute to aging in everyone, or whether it's bey trying to just slow the pace of aging by addressing some of these risk factors that may be environmental or some of them may be due to genes.

THEO VAN ERP: I think one low-hanging fruit may be to focus also on common factors that are associated with accelerated brain aging, like Hilleke Hulshoff Pol mentioned things like exercise, healthy lifestyle. These may actually -- these are hard to do for healthy adults sometimes, but they're even more challenging if you have a brain disorder that affects motivation or takes up a lot of, as mentioned earlier also, a lot of your energy to then also like pick up these generally healthier lifestyles that are in general good for healthy aging. I think maybe something that the -- maybe just a general focus that could help, not only in schizophrenia, but other brain disorders as well.

LISA EYLER: I was going to say that because those things are so difficult at an individual level, I think we need more systemic changes. So I loved that you started, Hilleke, with the picture of that lovely area for walking and biking that you have there at Utrecht. So how can we make the environment, the lived environment, particularly the places where people with schizophrenia tend to live in our communities, how can we make that more amenable to walking, biking, to exercise? How can we provide services that help people to stop smoking, to have a more healthy and nutritious diet, and so I know that this is something that we have started to study in the context of the board and care facilities here in California and I think those are -- we need to think about systemic solutions as well as individual solutions for this.

HILLEKE HULSHOFF POL: I think that's wonderful. Also, if you look at, obviously static biking is also beneficial to all, but it's very nice if they're available and that also brings back the brain to body combination. In fact, I also see some points whether it's genes or environment. For instance, the genes that seem to be implicated in these changes overall, right, not necessarily, seem to be implicating metabolic processes. So actually what I've heard earlier may be directly connected to that. We don't know yet. But that would be wonderful. Very hopeful.

ELLEN LEE: Thank you all for such wonderful talks. I was wondering if we could pivot a little bit, because one of the things that struck me from many of your talks was the recurrent description of these sex differences that we're seeing in some of these biomarkers and certain imaging finds but not others, and I was wondering if you could comment on some of that, given some of the striking things we've learned about yesterday.

DILIP JESTE: That is a great point. Are there any sex differences that were found in these studies?

LISA EYLER: I was just going to say that I think that the sex differences that we've found so far are important to follow up on, because I think it begs the question of mechanisms and, again, whether there would be sex-specific recommendations for folks. So if we see some of the telomere lengthening may be more specific in women with schizophrenia, if we also see high sensitivity C-reactive protein deficits. So it may be that, again, in our quest for personalized treatment that we need to consider someone's sex and really understand whether that -- what that clusters with in terms of biomarkers and then what implications does it have for the treatments that might be most effective.

HILLEKE HULSHOFF POL: I think I would totally agree with that point, but I cannot add much at this now on the sex differences. Most of the studies were basically underpowered to do so. But that doesn't mean, it's very important and we should probably try to look better into ways of being able to do this better.

SOPHIA FRANGOU: I have two questions for Dimitrios if possible. So the first one is about the nature of the EVs and the nature of the cargos of these EVs and their function. So there is some suggestion that the cargo of these EVs is not random, but there are some rules that are not clearly defined at the moment, I think, that determine what gets put into EVs and sort of thrown out of the cell. So that's kind of my first question, if you can say something about the nature of the cargo and whether the cargo itself sends messages to other -- that can influence other cells. So whether EVs of the brain carrying, for example, mRNAs for insulin, whatever, can influence other organs within the brain or outside the brain.

DIMITRIOS KAPOGIANNIS: It does, yes, they do. So the field of EVs in general has two ways of approaching the biological significance of the EV cargo. One is that they represent a disposal mechanism for unwanted material that the cell cannot process effectively, and we do that in studies where if you try, for instance, to block a lysosomal function that normally would metabolize a macromolecule that you no longer need, then you see an increase in EV production. So that clearly says that if a cell cannot deal with something effectively through its mechanism for proteostasis and whatnot, it packages in the EVs and makes it somebody else's problem.

So that is one end of the spectrum. The other end of the spectrum is clearly some signaling studies where they have shown uptake in the EVs released by one cell, they are uptaken by nearby cells, and their behavior is modified. This is very well documented.

So EVs can do both. Depending on the particular circumstances and the environment of the cell and the challenges that the cell faces, they do both, and it makes sense that they do both. If there is a -- if one neuron is under metabolic stress, there's a good chance that the other neurons are or will soon become under metabolic stress. So the increased trash, garbage, or what not, could be perceived by other neurons and they can be ready to face the same problem. So that's my brief. They are both trash and signals.

SOPHIA FRANGOU: And my second question, which is like out there: so you have showed and others have shown that there is some problem in sort of energy production and related to essentially lower enzymatic activity of the relevant kind of kinases. Would you envision a treatment where we create EVs -- we create the enzymes and we inject them? They would cross the blood-brain anyway.

DIMITRIOS KAPOGIANNIS: Absolutely. This is not science fiction at all. This is actually -- there are ongoing therapeutic efforts. Not for psychiatric disease, to the best of my knowledge, the whole field of EVs is just as big in therapeutic development as in diagnosis and biomarkers, and there is two schools of thought. One is that like we create EV-like particles, like you can think of the mRNA vaccines where there are liposomes, packed in liposomes, and you can decorate them with some EV markers that can facilitate their uptake by neurons, their passage through the blood-brain barrier and other barriers, or you can harvest naturally occurring EVs and then enrich them or take out some bad molecules, get some good molecules in. These are like -- the biggest challenge, I would say, is that standardization and homogenizing; to become a drug, it has to be greater predictability. So proof of principle, I think that we're very close as a field. You can show, especially stem cell EVs are like there are -- I have to review so many publications all the time, stem cell EVs for this and that. The hard thing is actually to standardize their cargo and to make it thus more of a drug.

SOPHIA FRANGOU: Analyze it, like the dose.

DIMITRIOS KAPOGIANNIS: Yeah, like it is one potential it would be even to take EVs from one person and gently engineer them and then introduce them to the same person, like auto blood transfusion. That would be also something neat, because it would bypass a lot of problems with developing biologicals as treatments.

SOPHIA FRANGOU: My last question: people that have done this thing, do they show that you need to get injected daily for example? Or is there a lasting effect? We don't know.

DIMITRIOS KAPOGIANNIS: We don't know. That is a gazillion dollar question, and it would be a lot better of course if a short course of treatment would be enough.

DILIP JESTE: Thank you. Lisa, you have a comment?

LISA EYLER: I was going to say that there have been a couple of questions in the Q&A that have to do with specificity, and I think all of us talked about how a lot of these findings, first of all, may not be specific to schizophrenia versus aging. So some of the things that we're seeing may just, again, be advanced aging as opposed to a different rate or process going on in schizophrenia. But also whether it's specific to schizophrenia versus other disorders.

And it was brought up that a lot of the things that we're seeing in the brain may be related to immune and metabolic dysfunction that can be there in many different disorders. So we could -- we don't know whether it's specific to schizophrenia or other types of mental disorders, or if it's also just something that we would see with diabetes or obesity, things like that. So I think where our field needs to go is to do some studies where we look across disorders and compare and contrast so that we really understand are there particular patterns of aging that are happening in schizophrenia that we can really target for those patients or are we looking at more general things that could be helpful for anyone, whether they have a major brain disorder or not.

DILIP JESTE: That is a very important point about lack of specificity. There are so many findings that initially get reported as where if you compared people with schizophrenia and healthy subjects and you find a difference, then sometime whether you look at other serious mental illnesses, there's bipolar, major depression, you have similar findings. So several of these are probably general findings for serious mental illnesses.

At the same time, I agree with Lisa that probably we really need to look at the nuances of the difference, that another issue is of course heterogeneity in schizophrenia that was also mentioned, that not all people with schizophrenia will have the same abnormalities. So on the one hand, it's a nonspecific finding that applies to general number of mental illnesses. On the other hand, it is specific to a subgroup of people with schizophrenia and what is that subgroup? So these are the questions to follow up.

I think we are approaching the end. Let me ask Ellen and Sophia if you have anything else to add.

SOPHIA FRANGOU: Nothing other than to say this was really a great session.

ELLEN LEE: Agreed. And then just a side note, I think that one of the things to address with specificity and this heterogeneity issue is going back to what Theo talked about, is building a large consortium of data that allows us to harmonize and look at large populations of people, and a lot of our initial assumptions may be disproven or we may find new data-driven hypotheses. So I think that's something we want to talk about more at the end of the synthesis. So thank you all.

DILIP JESTE: Thank you all. This has been a great session. I know I learned a lot. So thanks to all the speakers, co-chairs, and everyone.

We have a 15-minute break and we will come back at 2:15 East coast time.


CAROL TAMMINGA: My name is Carol Tamminga, and I'm the session moderator of the last session. That allows me to say welcome to everybody for the final session, and also, just before we start, I want to say thank you to the NIMH and to the meeting organizers, Dr. Lee and Dr. Frangou, for really putting on such a lovely great list of sessions, maximizing the amount of information that we get, but also the discussions that can get, that we can have around these sessions, too.

It's hard in a meeting like this to say that we have saved the best to last, but I say if you look at the list of speakers that we have in this last session I think it's really a knockout. Clearly, it's the A team, from my point of view.

First we'll have Dr. Anissa Abi-Dargham, she's the chair of psychiatry at Stony Brook, and she heads up the imaging program there. And she's going to talk about the opioid receptor and the D1 receptor, and the people in these talks will talk about the topic of treatment targets, interventional targets, and the development of those.

And we have Dr. Sagnik Bhattacharyya, and Dr. Bhattacharyya is from the Institute of Psychiatry at King's College in London, and she's going to talk about a topic that her site has done a lot of research about, and that's about cannabidiol as a treatment targeted intervention.

Then we'll hear from another international person, from Canada, Dr. Tarek Rajji, and Dr. Rajji is from the University of Toronto, from CAMH, and he will talk about neuromodulation, which is really is a specialty in CAMH.

We'll end with Dr. Phil Harvey, who is our own U.S. researcher that needs no introduction. He's from the University of Miami at the Miller School of Medicine, and works a lot as the head of research at the Miami VA, and Phil Harvey is going to talk about cognitive rehabilitation.

I'm going to ask each one of these speakers to introduce the next speaker. Dr. Abi-Dargham to just turn it over to Dr. Bhattacharyya, to turn it over to Dr. Rajji, and then on to Dr. Harvey. And then at the end, we'll have a question and answer time at the end.

With that I'll ask Dr. Abi-Dargham to start her talk to start off this session. Thank you very much.

ANISSA ABI-DARGHAM: Thank you, Carol. Also, thank you for giving me a promotion I wasn't aware of. I'm vice-chair for research, not chair. I also want to thank the organizers for this fabulous two-day program. I've really enjoyed the talks of all the speakers, but also a lot the discussions.

These are my disclosures here. I was asked to speak about our work in the development of D1 as a target for therapeutics in schizophrenia and our initial beginning work with the kappa opioid receptor, potentially also as a target for treatment.

You'll forgive me for these being pretty much a work in progress kind of presentations.

My interest in the D1 and the kappa opioid receptor derives from the following. My group, as well as many groups around the world, have been looking at dopamine transmission in schizophrenia for very good reasons. There's no doubt that dopamine is involved. But there are now two types of findings that have emerged.

On the one hand, in the striatum, and especially the associative striatum, there's an increase in dopamine that relates to psychosis and other related measures. But then outside of the striatum, extra-striatal regions, there is a deficit of dopamine.

Two implications -- the first one, the obvious one, if there's a deficit of dopamine, especially in cortical regions, this really brings attention even more so the D1 receptor, which is the main mediator of cognition for dopamine in the cortex, and stimulating the D1 would be one reasonable thing to try.

The second consequence is that having an imbalance in dopamine, where one area has too much and other areas have too little, brings up the possibility that there could be local factors modulating dopamine, in opposite manners. We don't know what it is exactly, but it shifts the attention to something else, not primary to the dopamine system but maybe modulators of dopamine, and obviously the kappa opioid receptor system is one of those. There are many others. Yesterday Sophia mentioned our interest in the cholinergic system, but we won't talk about this today.

So the D1, I'm going to start with that, it's been known from nonhuman primate studies and other animal studies that you need an optimal level of stimulation of the D1 receptor to have optimal working memory performance. So aging and Parkinsonian monkeys would benefit from an agonist, while stressed and amphetamine-treated animals would benefit from an antagonist.

The D1 receptor is present on the basal dendrites of pyramidal glutamatergic cells, and also on GABAergic interneurons. Those glutamatergic cells are known as the delay cells, because they fire during the delay period of a working memory performance, and the actions of D1 on those, the GABA and the glutamatergic cells, finetunes the cortical circuitry to be able to subserve cognition and to have optimal working during working memory tasks.

Amy Arnsten has explained very nicely the cellular components of this inverted U-shaped curve. So when you have D1 stimulation, two things are important. Many things happen, but two are important. The NMDA receptor gets phosphorylated, and is inserted into the synapses, transcends these connections, but also the D1 stimulation will induce opening of the nearby potassium channels to kind of reduce the spreading of the firing to neighboring neurons.

So initially there's a little bit of an overshoot in this kind of stimulation. As the potassium channels open, you have kind of a finetuning of the signal. But then if there's too much stimulation, it spreads almost too far and there is nonspecific suppression of (inaudible) so that could explain why it's only a certain range of stimulation that works best.

Obviously, this is not new. We've known for a long time from postmortem imaging studies there's a deficit in tyrosine hydroxylase immunolabeling in schizophrenia in the prefrontal cortex, from CSF studies, that dopamine metabolites are inversely related to cognition. Clinical studies have shown that cognition can be enhanced with amphetamine, increasing dopamine. And imaging studies have shown that blood flow can be improved by amphetamine or DAR100, which is a D1 agonist.

The most conclusive evidence, though, came from molecular imaging studies, our own and that of others. We have used an amphetamine paradigm and shown that patients with schizophrenia on average have no dopamine release, the inability to displace the radiotracer as pictured here, compared to controls. The group of Romina Mizrahi showed similar findings, but using a cognitive challenge task. We also showed that dopamine release capacity in the prefrontal cortex is related to the ability to activate the prefrontal cortex during working memory performance.

Now I'm going to turn to the challenges in developing D1 targeted therapeutics, and the first challenge is very understandable. It's the U-shaped curve itself. For that to be optimally targeted, one has to know where is the individual on this spectrum to be able to address this. That's a difficult task, and the difficulty -- there are many factors that play into that, and I'm going to walk you through those.

The first one is D1 expression itself. PET imaging studies have shown very inconsistent results. Some have shown decrease, some an increase. We've shown an increase that then with a second study we found to be only in the drug-naive patients. This is consistent with data in nonhuman primates from the Lidow group, showing that D1 expression in the prefrontal cortex changes with antipsychotics, with D2 blockers, over time.

So what I think is happening is that there is a heterogeneity in D1 expression in schizophrenia that could be related to exposure to antipsychotics, age; D1 receptors decrease with age, we've shown that, and chronicity.

The second contributing the challenge is to know the level of occupancy of the D1 receptor by dopamine. If we extrapolate from what we know about the occupancy of the D2 by dopamine, they have roughly similar affinity, we can basically infer that it's set to be very low, it's 10 to 15 percent. But this number is really valid as a kind of an integrated type of occupancy over time. We don't know what is the dynamic range of occupancy when a person is undergoing a cognitive task or stress or physiological activities.

The third is how much to stimulate. In nonhuman primates, it's been shown that this dose has to be very little, could be very little, infinitesimally little, actually, look at these numbers. That enough of this small dose will reverse the cognitive impairment that results from chronic treatment with haloperidol. That has to be shown to be the case in humans.

Positive allosteric modulation or partial agonist is the other question. Positive allosteric modulators will enhance the signal that comes from dopamine itself. The caveat here is that if there is a profound deficit in dopamine, that may not work. Partial agonist will provide a blanket kind of stimulation at whatever intrinsic activity and occupancy the drug is producing, and that is good. We have some stimulation. But it may not be the way dopamine should work, to have just constant stimulation that is unvariable, with a situation the subject is going through.

Finally, the availability of drugs. What we've had up until now are catecholamine-type drugs, and these are very difficult to deal with because they have poor oral bioavailability, fast clearance, they produce tolerance, peripheral side effects of hypotension, which limit the dose, which means that we've had to do studies with almost no occupancy because the drugs at the doses provided we know did not really penetrate into the brain.

What is the opportunity? Now, fortunately, Pfizer and David Gray, the chemist that was a Pfizer, now at Cerevel, has developed a new series of compounds that are not catecholamine-based. Because of that, they have better pharmacokinetics, better oral bioavailability, better blood-brain barrier penetration, and they can produce higher occupancy, if we know what occupancy we want, with fewer side effects.

This drug, PF-2562, I'll refer to it, renamed as Cerevel 562, I think, currently. This is a drug that we're actually starting to work with. The affinity at D1 and D5 are nonnegligible. The intrinsic activity is slightly better at D5 versus D1. Initial testing in Parkinson has shown that it's well tolerated and may give a signal on motor skills. Testing in nonhuman primates, with the drug applied iontophoretically in the prefrontal cortex, increased firing in the DLPFC in aged monkeys, so that's great proof of concept. And initial study in patients with schizophrenia showed improvement in all groups, including placebo.

This is a drug that now is what we will use in this study, we call the TRANSCENDS study that is multisite, in collaboration with Cerevel, between Yale, the lead site, Stony Brook, Columbia University, and UPenn. We will give acute dosages on five separate days, five separate doses; we look at fMRI biomarkers during a spatial working memory task that has been very well validated from all the animal studies examining this. So a translationally valid task.

We will focus on early phase of the illness because of all the changes with D1 expression and dopamine over time. If we add on to antipsychotics, we will avoid the ones that have D1 affinity. And basically this is a study that is now ongoing, and we hope that we will have interesting findings to report to you. Obviously there will be a lot of next steps. If we find a signal with one of the acute doses or more than one of the acute doses, this will obviously be something to take forward into subchronic and chronic designs. We will need to link the computationally modeled fMRI outcome measures to the cognitive effects and functional outcome.

It would be nice if we had predictive biomarkers to be able to tailor to specific populations. There are some that one could think about. And it's possible that we would need to think about playing with the selectivity for D1 versus D5.

Now I'm going to turn over to the kappa opioid receptor. The interest in this receptor is because its presence throughout the brain, and some very key regions like the midbrain and dopamine cells, the GABA cells, that project to the dopamine cells, the striatum, the medium spiny neurons, both the D1-bearing and the D2, and can affect the balance of these, and also in the cortex. So all the key regions that we know are involved in schizophrenia.

It is known to modulate dopamine in complex manners. It also weakens the D1 versus the D2 medium spiny neurons, so basically can contribute to anhedonia type of outcome or psychosis type of outcome. Both of these are kind of involved in these symptom domains.

It also modulates GABA, glutamate, serotonin, and so the relevance is really far beyond just one symptom domain. This is something that is potentially very interesting.

Pharmacological studies have shown that kappa agonist, and in particular one that people know a lot about is salvinorin A, can produce psychosis. And that can be reversed by pan-opioid antagonists such as naloxone, naltrexone. We performed a meta-analysis, including 30 studies, 400 patients, of the available literature, which included basically pan-opioid antagonists, and buprenorphine, which is a mixed kappa antagonist mu opiate agonist. And what we found is that there is a therapeutic benefit on positive and negative symptoms, even when added to antipsychotics, presumably through the kappa antagonist mechanism. It's a small effect size, but obviously these were just not necessarily optimized in any way.

Just to kind of make it clear, these drugs bind with some affinity to all three types of opioid receptors. Quite a bit to kappa, but also to mu and a little less to delta. So the reason we are suggesting that it's all mediated via the kappa, potentially, is because while some are antagonists at mu, one of them is an agonist at mu. So we think that the mu is probably not contributing to the therapeutic signal. But obviously this has to be confirmed.

In the meanwhile, we've started to explore the kappa opioid receptor system in schizophrenia. We have a small study to look at only ten patients, ten controls. And these are some of the initial images. You see that it's detectable throughout the brain. We're using an antagonist radiotracer developed by Henry Wong at the PET Center.

Just to show you a little bit the distribution, amygdala, uncus, insula, some of the highest regions, as well as other cortical regions. This is the ventral striatum. It's also pretty high, and the associative striatum. These are all the regions of interest for us, and also other cortical regions. This receptor binds everywhere, even in the cerebellum, so there's no reference region here, and that's why we're reporting the volume of distribution.

We will relate the kappa opioid receptor measure to neuromelanin sensitive MRI, which is a measure we have developed while I was at Columbia a few years ago, validated a way of using MRI to examine the deposition of neuromelanin in the substantia nigra and VTA, and showed that it actually relates very well to the gold standard method, which is the PET measure of dopamine release in the striatum. So as a cumulative integrative kind of index of dopamine availability over time.

Opportunities and next steps. The KOR has untapped potential in schizophrenia as therapeutic target for potentially more than positive symptoms. The KOR levels could be used to stratify patients if we find a positive signal in our imaging study and replicate it. There is actually a potential drug, a J&J drug that was used in treatment of anhedonia and showed good results. We even reached out to Johnson & Johnson and asked them if we could collaborate on that. But there may be other drugs that would be available to test this.

With that, I want to thank the TRANSCENDS team that is just really an amazing group of people to work with on the D1 development, and my own lab members enrolled in the kappa study, the funding agencies, and the patients and families who volunteered. We couldn't do anything without them.

And with that I want to turn it over to Dr. Bhattacharyya.

SAGNIK BHATTACHARYYA: Thank you. Firstly, thanks to the chairs of this meeting, Dr. Lee and Dr. Frangou, as well as the organizers, the NIMH team, led by Drs. Rowland, Wijtenburg, and Dr. Fisher, for inviting me. It's a great pleasure to be here today, and very fascinating set of talks.

I'm going to talk about cannabidiol. Perhaps it will come as no surprise to this audience that cannabidiol is the panacea for all the ills of humankind. And definitely for that reason, we are talking about it today. But let's for a moment climb down from that lofty ambition of solving all the ills of society using cannabidiol to focusing really on kind of its potential as an antipsychotic in older adults.

Let's look at the current evidence. But before we do that, I think it'll be worth briefly taking about what cannabidiol is. Although I doubt that anyone in this audience will not know about cannabidiol, but I think it's often conflated with cannabis or medicinal cannabis, so I think it's worth noting that cannabidiol is one of several cannabinoids present in the extract of the cannabis plant. The two that we talk about most often are delta-9-tetrahydrocannabinol, or THC, and cannabidiol, or CBD. And I'm going to focus more on pure CBD effects in the course of this talk.

And we know that different strains of the plant have different proportions of these cannabinoids. So what do they do? We know that THC is the intoxicating bit of cannabis, and it can induce psychotic symptoms in healthy individuals, worsen them in people with schizophrenia, while CBD is the nonintoxicating bit of cannabis. It definitely does not induce psychotic symptoms and may in fact have some therapeutic effects in terms of anxiety and potential antipsychotic effects.

The question that arises is why cannabidiol? Why are we interested in that? So let's go back, a few years back. When we started some of these studies we knew that THC had acute psychotomimetic effects, while CBD did not have any such effect. So we wanted to directly compare and contrast their effects on behavior and brain function. What we did was we gave a group of healthy individuals a small dose of THC or a small dose of CBD, on different occasions and compared that with an inactive pill on a third occasion.

And we looked at what was happening in terms of -- in their brains. In these individuals, THC caused acute psychotomimetic effects, and CBD did not have any such effects. But when we looked at what was going on in the brain, we found that, as you can see here, the effects of CBD are shown by the blue bars, the effects of placebo are the green bars, and the hatched bars show the effect of THC.

The effects of CBD and THC, relative to the effects of placebo, across a number of functional MRI paradigms that we use to look at their effects on brain function, were in the opposite direction. So, memory, anxiety, response inhibition task, and visual and auditory stimulation tasks, suggesting that -- and also an during attentional salience processing task -- we found that the effects of CBD and THC on brain function seemed to be in the opposite direction relative to the effects of placebo, suggesting, kind of making us wonder, that does CBD then block some of the acute psychotomimetic effects? Which led us to do another smaller study, where we gave a small number of healthy individuals THC on two occasions, and on one of the occasions the THC was preceded by a small dose of CBD. On another occasion, there was a placebo pretreatment.

What we found was that THC preceded by placebo pretreatment was associated with an increase in psychotic symptoms, as shown here by the green line. And these increased psychotic symptoms were blocked by a certain extent by CBD pretreatment in those same individuals, suggesting that CBD could block some of the acute psychotomimetic effects of THC.

Following on from that, we have now got a number of trials that have really looked at the potential of CBD as an antipsychotic, given that we've found that CBD blocked some of the effects of THC and opposed some of its effects, and THC having psychotomimetic effects, then, it would be reasonable to think that CBD might have antipsychotic effects, and yes, Marcus Leweke found that CBD was not inferior to Amisulpiride, an established treatment in acute schizophrenia. Subsequently, GW with their Epidiolex CBD formulation found that CBD was better than placebo in terms of reducing the psychotic symptoms in people with chronic psychosis.

However, a subsequent study by the D'Souza group found, which really focused on effects of CBD on cognition, particularly memory, they didn't find that CBD was of any use in terms of psychotic symptoms. There are very many reasons why they might not have found such effects, but perhaps we can, in the interest of time, we can move on and we can come back to this if anyone is interested.

So the evidence suggest that CBD might have some antipsychotic effects, so we wanted to look at this in a group that really has an unmet treatment need, people at clinical high risk of psychosis. Many of you might be familiar with this concept, but for those who may not be very familiar, these are people who have low-grade psychotic symptoms that are qualitatively similar but less frequent, less severe, and less persistent than those experienced by people with acute schizophrenia. And they're at very high of transition to frank psychotic disorder.

What we wanted to know was that did CBD have a potential as an antipsychotic in this group of people. So what we did was, we did a double blind parallel arm randomized control study where we gave a single dose of CBD or placebo to people with a clinical high risk of psychosis and studied them using functional MRI. We also studied a group of healthy individuals who did not receive any drug. And what we found was that during a verbal learning task inside the scanner, the healthy individuals had -- the task engaged these parts of the brain shown by the red blocks, bilateral striatal areas as well as midbrain extending to the medial temporal cortex. And in these regions, in people, clinical high risk of psychosis people, who received placebo, that there was reduced activation in these regions, while in people who received a single dose of CBD, the effect was such that it seemed like there was intermediate effect. So their engagement was intermediate to what we saw in the healthy controls and in the placebo-treated clinical high-risk patients.

Suggesting, perhaps, that even a single dose of CBD might be trying to ameliorate some of the neural abnormal activation may underlie psychosis in these individuals. And then we saw a similar pattern when we looked at a different cognitive paradigm. This was a fear processing task in the same group of individuals. Again, a similar pattern of effects. And then, however, because these were parallel arm design studies, we could not be certain that it was CBD that really ameliorated some of that abnormal activation. So we did a within-subject design study in people with established psychosis, where we gave them CBD or placebo in a double blind randomized control design. And again we found that relative to healthy controls, there was increased engagement of some parts of the brain, such as the medial temporal cortex and the prefrontal cortex, while following CBD treatment, some of those abnormalities seem to be attenuated following a single dose of CBD.

To summarize, what we found was that CBD in healthy individuals seemed to have some effects on brain function and psychotics in terms that were opposite to those of the psychotomimetic effects of THC and its neural underpinnings. And in clinical high risk for psychosis patients and those with established psychosis, a single dose of CBD seemed to target brain regions implicated in schizophrenia, and its effects seemed to be consistent with its antipsychotic and antianxiety potential.

Of course, there is evidence of antipsychotic potential from two out of three RCTs in patients with acute or chronic psychosis. And to add to that, most of these RCTs showed that CBD was very well tolerated in people with psychosis. So we then looked at the long-term effect of CBD, longer term in the sense that sustained treatment, because this is unpublished, I'll just present a bit of the data. The key take-home message is that we found that three-week treatment with CBD seemed to cause a greater reduction in some of the clinical high-risk for psychosis symptoms in these individuals, as well as causing reduction in the distress associated with psychotic symptoms in these individuals.

Let's change tack from there. It seems that there is some evidence in terms of antipsychotic potential for CBD. It seems to really target some of the key brain regions that we know are functioning abnormally in people with psychosis. What are the essential characteristics? What are the key characteristics that we would like in antipsychotic that we want to use in older adults?

Of course, efficacy, it has to work as an antipsychotic. It has to reduce the psychotic symptoms for it to be useful. And the other issue is, and this is true for all antipsychotics in all age groups, that it has to have a good safety and tolerability profile. But this is a particular interest in people, in older adults who may also have a number of comorbidities, physical and other comorbidities, for all of which they might be receiving a number of treatments. So there is a particular need that any treatment that we add to whatever they're already receiving really needs to be very well tolerated, really need to, perhaps, not interact as much, causing more side effects.

So we looked at the effect, the safety tolerability profile of CBD in older adults, and again, I want to remind you that there are no studies looking at CBD for psychosis in older adults, so we did a meta-analysis looking at the evidence regarding the tolerability of CBD in older adults, that is, people aged 50 and over. There are very few studies, as you can see here, and most of them are crossover studies from which there was any kind of detectable, any reported evidence, and very small sample sizes, from 6 to 15 subjects. And again, really, two studies, and not in psychiatric indications. One in Huntington's disease and another in intraocular pressure. Again suggesting that CBD seemed to be no different from placebo in terms of its tolerability profile. So that's good news.

Then, what do we know now? We know that CBD, from the limited evidence that we have, seems to be very well tolerated. But what about its efficacy? Is CBD effective as an antipsychotic, do we have convincing evidence for that? Of course, we do not have at this point of time any definitive evidence that CBD is effective as an antipsychotic in any age group. Of course, there are no studies in older adults, so that's yet to be done.

But then there are particular issues that we don't know, that need to be answered in this population. So will CBD be safe and tolerated well in older people with psychosis? So that's a particular issue in this age group. What is a safe dose of CBD to use in older patients? Are there going to be clinically relevant drug-drug interactions between CBD and any other treatments that people may be on? And how might CBD work? So these are all unanswered questions that perhaps need addressing in the next years or so.

What else is going on? I thought that I'll kind of refer to a few studies that we are doing that might help address some of these questions. For example, we have an upcoming multicentered study in people at clinical high risk of psychosis, looking at 300 patients, half of whom will get CBD and half will get placebo, as an add-on to their treatment as usual, over a six-month period, to look at any efficacy in terms of psychotic symptoms in this population. We will also neuroimaging in a subsample of these individuals, in 100 participants, to try to understand at a systems level, which brain areas does CBD target and/or what kind of targets might be related to its beneficial effects in these individuals. So that's one study which will perhaps speak to the efficacy question, but not so much the specific aspects in older adults.

We have another study underway at the moment in people with Parkinson's disease psychosis. While this is quite different from nonaffective psychosis, unrelated to other indications, but it still is relevant in the sense that these are also older people who have psychotic symptoms and they also receive a number of treatments. So give some indication as to the safety, tolerability aspect as well.

And we have just finished the dose-finding study. We have tested a number dose levels from 200 to 1,000 milligrams. It's a little bit too early for me to present any results. All that I can say is that CBD seemed to be very well tolerated; all the doses were tolerated. We expected that perhaps over 600 to 800 milligrams may not be tolerated, which was not the case. All the doses were very well tolerated, and there were no serious adverse events, no withdrawals during this dose-finding open label study. And we are in a couple of months we are going to move to the double blind randomized control phase, using this dose that we have identified. We have yet to make a decision on the dose. As soon as we make that, we expect to start this study from around April, May, as soon as we identify the dose. Of course, we also will do a neuroimaging study to try to understand some of those mechanisms in a subsample of the participants.

And then I have only talked about two trials, but we have a number of other trials, some of which are international multisite trials, led by my colleagues, but I'm also part of those trials, in other stages of psychosis, which I thought may not be as relevant because most of those are in younger adults. But those are again, studies that will speak to the efficacy and safety evidence for CBD.

All of this could not have been possible but for the support from my colleagues as well as patients and volunteers, and my funders.

I'll stop there, and I'll hand it over to the next speaker, which is Dr. Rajji.

TAREK RAJJI: I also add my thanks to the NIMH and the organizers and the chairs for a terrific workshop, a very highly needed workshop.

I'll be talking in the first part of my talk about neurophysiologic target or potential target for neurostimulation interventions, and the second part I'll talk about potential neurostimulation interventions that could engage that target to enhance cognition in patients with schizophrenia.

So that target is based on using EEG and it's a cross-frequency coupling measure called theta-gamma coupling whereby the amplitude of gamma oscillations are modulated by the phase of theta oscillation lower frequency oscillations. Based on early computational and preclinical work, the model is that, in a simplistic way, gamma oscillations represent items of information being held during working memory, and the coupling could code for the ordering of this information during the working memory task example.

The task that we've been using to study working memory and study theta-gamma coupling during working memory is the n-back task, which as many of you may be familiar with, there are different conditions. For example, on the 1-back condition, the person is asked to decide each letter that they see whether it matches one letter back or not. Then there are 2-back and 3-back conditions.

If we focus on the 2-back condition, so if we look at the second B letter, we can see it matches two letters back. So that's what we call a target trial. Unlike the K letters during what I'm showing here as an example, which are a non-target trial, because they don't match two letters back.

However, the two K trials are distinct, because the first K trial does not match any letters before it, unlike the second K trial to the right which matches three letters back, and so it's close to the 2, and that's why we call it the non-target lure. The reason why I'm mentioning this point is because when we first started to study theta-gamma coupling in healthy individuals, what we hypothesized is that to answer correctly a target trial like the B trial, it's important to remember the letters that were seen before on a screen, but it's also essential to remember the order in which these letters were presented to answer correctly a target trial, and therefore coupling would be expected, which codes for the sequence for the order of information would be expected to be high.

Unlike a nontarget non-lure trial, where familiarity may be sufficient. So if I'm seeing the first letter K and I say to myself I haven't seen K recently, therefore I don't need to recollect the order of the letters, and I'm able to answer the letter correctly, the track correctly, and then for a nontarget lure trial, familiarity may be sufficient if I can limit my familiarity to the previous two letters, but if I'm able to recollect the order of information, then my chance of responding correctly to a nontarget lure trial also increases.

So the prediction was that -- and what we've shown -- is that the theta-gamma coupling here measured during modulation index of the gamma amplitude by theta phase is highest during, across different conditions, highest during the target trial, lowest during the nontarget non-lure trial, and then in between on the nontarget lure trials, and you can see that especially on the 2-back and the 3-back. On the 3-back, there are two types of lures where one of them is quite distant. So that's the lure plus 1. So that's four letters back becomes quite distant to really benefit from collecting the order of information.

So this study was done in a group of healthy individuals. Some of them were older. Then we looked at this measure and working memory performance in individuals with disease conditions here, older healthy individuals versus older patients with early Alzheimer's disease or MCI. What you see on the lower left side on the bar graphs is that even though patients with MCI were not that impaired on performance of the working memory task, the 2-back condition specifically, compared to healthy individuals, of course the Alzheimer disease patients were quite impaired. However, the MCI, despite the fact that they were not impaired, they were quite impaired on the theta-gamma coupling measure compared to healthy individual, suggesting that maybe the theta-gamma coupling index is representing a cognitive reserve process that needs to decline below a certain threshold to start manifesting behaviorally.

We also found that there is a strong correlation between performance on the working memory task and theta-gamma coupling across the whole sample, but even if we exclude those who are the Alzheimer disease group because they were quite impaired on performance.

So then we asked the question, so the data I showed you is that we were measuring theta-gamma coupling during the performance of the working memory task that requires order of information, but if this measure is indexing the ability to order information in general, then we asked the question does it associate with performance on other cognitive tasks that are done not during the EEG collection but a few weeks before or after, and so we tested this question. We tested the hypothesis in individuals who performed before or after the n-back session, paper and pencil neuropsychological test, the top two that you see on the screen, the PASAT and the Trails Making B, require ordering of information in certain sequence, in contrast to the three tests that you see on the bottom, Boston naming, Digit Symbol Coding, or Judgment of Line Orientation, that do not require ordering of information. We predicted that theta-gamma coupling during the n-back task would be associated with performance on the first two tests but not the last three.

This is the sample that we used to ask that question. It's again an older sample of older individuals who are healthy individuals or people with depression, but in remission, or people MCI with or without depression, again history of depression, these are not acutely depressed.

To note also the diagnosis did not really matter in terms of the results here, and we saw the pattern as we predicted it, so as you can see in a multivariate analysis, MI again modulation that's of coupling was associated as we know with theta-gamma coupling, with the performance on the working memory measure, which is the 2-back, but also on the other one, the PASAT and the Trails B, but not on the other tests that do not require ordering of information, suggesting that it's an index -- the theta-gamma coupling is indexing ordering information independent of the task that is being related to.

Then we asked the question, does change in theta-gamma coupling, would it be associated with change in working memory performance? So here we took attention of two cohorts, so independent cohorts, one is a group of older adults with again major depression, but in remission, who are tested at three timepoints with several weeks in between, and a second cohort of older healthy controls tested twice with about 12 weeks in between.

And those were tested using the n-back task while collecting also EEG during the n-back task. What we see in both cohorts, independently, that there is a strong association between change in theta-gamma coupling at the individual level and change in performance, and here we are using the 3-back condition.

What about patients with schizophrenia? So we had published earlier that patients, these are adult patients, not older individual, but adults with a mean age in the 30s, are impaired on theta-gamma coupling during the n-back task across the different conditions, different n-back conditions, all the way from 0 to 3-back.

So can we intervene, to change the theta-gamma coupling, to engage it and moderate it and potentially have a positive impact on cognition and working memory? One brain stimulation TMS-based paradigm that we have been using in our team, with our team, is called paired associative stimulation, which simulates spike timing dependent plasticity. It was the first developed for the motor cortex where peripheral nerve stimulation of the median nerve is paired to TMS stimulation to the cortex here in the motor cortex.

We had adapted this paradigm to look at plasticity changes or LTP-like changes to the prefrontal cortex by combining TMS with EEG and the output in this case would be cortical volt activity overlying -- using EEG overlying the DLPFC. So this in healthy controls, this is the first study we published in healthy controls, which showed that this paradigm can result in change in plasticity over the prefrontal cortex compared to a control PAS intervention, but also what we saw is a robust increase in theta-gamma coupling, this theta-gamma coupling in response to the TMS pulse after active PAS compared to control PAS.

And then we adapted this -- we tested this paradigm in patients with early Alzheimer disease and again, here we see that patients with early Alzheimer dementia, they have impaired plasticity over the DLPFC using this paradigm, even though it's still present, but it's impaired compared to healthy controls, and then we randomized -- this was a pilot study that now we're running a larger efficacy trial based on these pilot results in patients with MCI, but in this pilot work, then we randomized these people, 32 patients with Alzheimer dementia, to receiving a two-week course of active PAS versus control PAS, every day five days a week for two weeks.

What saw is that the patients who were randomized to the active arm they do experience increase in plasticity on the day after the intervention. But they also experienced improvement in their working memory task using the n-back, and there was also an improvement in theta-gamma coupling during the performance of the working memory task. You can see that the profile of the findings, they are comparable. Unfortunately, here we did not see persistence of the effect at 7 day and 14 days after the end of the intervention, suggesting a need for maybe additional sessions or boosters.

Also, in this population, these are only patients with Alzheimer dementia, we saw an association between performance on the working memory task and theta-gamma coupling.

There are other forms of brain stimulation that induce and engage -- have been shown to engage theta-gamma coupling. One is so far we talked about the TMS based intervention; this is transcranial direct current stimulation, where it's using a direct current of typically 2 milliamp current, and this is a study by Jones et al, showing that in younger adults using a tDCS of 1.5 milliamps for 15 minutes, alternating between the right prefrontal cortex and the posterior parietal cortex, that this paradigm can -- this intervention can strengthen theta-gamma coupling, the modulation again of gamma amplitude to the theta phase, but also the degree of strengthening of coupling is associated by enhancement of working memory performance. Again, these are young healthy adults.

Another form of electrical stimulation uses an alternating current stimulation, or tACS, in contrast to the direct current stimulation I just mentioned, and in this study, this group took a younger group of younger adults, healthy adults, versus older healthy adults, and the older healthy adults were impaired on performance on the working memory task and on theta-gamma coupling, but after they received active tACS -- and this was actually tACS individualized to the theta rhythm, the tACS engaged theta-gamma coupling, improved theta-gamma coupling, and through that engagement, performance on the working memory task improved in the older adults to become comparable to the younger adults.

I will end by showing you this slide on a trial that we just started to -- we're enrolling, will be enrolling 270 older individuals with severe mental illness, many of them will have schizophrenia primarily, and then we randomize to receiving cognitive remediation plus active tDCS or sham tDCS over a period of two to five years, and over time we'll be looking at -- with booster sessions. So they will get a course of eight weeks of tDCS plus cognitive remediation or sham tDCS plus cognitive remediation, with booster session. Then we'll be looking at cognitive outcomes, real-world functional outcomes, also be looking at balance, given the relation between executive function and balance. This stimulation would be targeted at bilateral post -- DLPFC.

And I will end here by thanking all the participants and the care partners who contributed to this work, large number of colleagues and large number of terrific trainees and learners who really led all of this work, and the funding agencies.

With that, I will pass it to Dr. Phil Harvey.

PHIL HARVEY: Thank you, Tarek.

I'm going to talk about cognitive remediation and considerations for older participants in the schizophrenia spectrum and also present some data we've been collecting looking at older healthy people as well as people with MCI. I think you'll be happy to see that the results that we're obtaining are very similar to what Tarek was just telling us about.

I have to disclose a conflict of interest. I'm chief scientific officer of i-Function, Inc. This startup was developed at UM and has licensed IP that we developed with NIH funding. Also, I receive royalties from the BACS, which is owned by WCG VeraSci that in another piece of evidence of the good judgment of the NIH was developed with NIH funding there, as well. These are conflicts of interest, but they wouldn't be existing if the NIH wasn't supporting our efforts to assess and treat cognitive functioning across the severe mental illness spectrum.

Cognitive remediation is something that for a while was seen as controversial, maybe because of the outrageous claims of certain cognitive remediation providers, but recently it's become very clear that the evidence is solid. Patients will engage when they can train at home. Train-at-home studies have found the same rate of training as come-to-the-office studies. Cognitive benefits are clear. But we need to understand the potential treatment targets and the potential range of benefits. We need combined therapy, and we need to figure out what the correct combinations are.

Basically, prior and newer meta-analyses have both suggested very consistently that greater functional gains result from targeted skills training in concert with computerized cognitive training. So it's definitely combination therapy has proven to be better.

Several randomized trials have also shown that combining skills training and cognitive training lead to more rapid and more proximal functional gains. This is a study that I did with Chris Bowie, who is now at Queens University in Kingston, Ontario. It's a randomized trial where people with schizophrenia were randomized to cognitive training alone, skills training alone, or the combination of the two, and what you see is that cognitive remediation improves cognition with or without the concurrent presence of skills training.

Skills training improves functional skills, measured by the UPSA, regardless of whether or not you get combined cognitive training. But the impact of cognitive remediation and skills training is higher or equivalent in the group that gets both, and everyday real-world functioning improved over the course of a 12-week period, detectably to blinded observers who could see that people were improving in their everyday functioning if they received the combined training but not either one of the monotherapy interventions.

Similar results have been presented repeatedly by Susan McGurk, who's at Boston University. She's delivered cognitive enhancement therapy to people with mental illness who don't respond to supported employment. What happens is, supported employment is a wonderful intervention with about a 20 percent response rate. What we want to do is to augment that response rate, and give people some kind of booster so they can get more benefit. What she showed is that if you took people who were previously nonresponsive and then you gave them six months of computerized cognitive training or an augmentation of their vocational rehabilitation, you get a big boost in employment outcomes, the percent competitively employed, the weeks of competitive employment, and those outcomes, the positive outcomes, continue to accelerate after the end of the formal cognitive remediation intervention.

Again, suggesting that work is probably the best cognitive training intervention that we can give to people. We just have to get them to the door and get them employed in the first place. The combination of vocational rehabilitation and cognitive training leads to about a 50 percent success rate, in contrast to the typical 20 percent success rate that you see with supported employment alone.

So the relevance to older participants, in the Bowie study, younger participants made greater training gains, but there were still significant gains for older participants. Several studies of very chronic patients, including people who were institutionalized and somewhat older, have found essentially equivalent gains in terms of the effect size for cognitive improvement, and have also shown functional gains, limited of course by the fact that people who are institutionalized can only work at an institutional-type employment setting.

A largescale and recent meta-analysis suggested that the signal for pre-morbid education was more important than the signal for age in terms of things that moderate the response of cognitive remediation.

Addressing very clearly the ability of older individuals to benefit from cognitive training, there was a very large-scale study called the ACTIVE trial that launched in about 2004, wherein 2,800 older people with a mean age of 73 were randomized to one of four training arms: speed of information processing, memory, reasoning, and a control condition. All training resulted in short term improvements in cognitive ability, but speed training and reasoning training led to persistent gains. The improvements were sustained at a 10-year follow-up. Let me show you these data.

This is the gain in speed training, persistent over ten years, associated with the individuals who were randomized to that condition. So what we see is that there is a big improvement. The mean speed to completion, in fact, was cut by 50 percent, and even after ten years -- and remember, if you start out when you're 72, you're going to slow down somewhat with aging anyway -- there's still a very substantial benefit relative to baseline.

Perhaps most important for all of us, the ten-year risk of dementia, as a function of training, was notably associated. If you received speed training for ten hours, plus four hours of booster training, two hours a year for the next two years, your risk of developing dementia in the next ten years was half of what was seen in the people who received the control condition. In fact, the importance of the booster training was seen even in the speed training intervention, so it does suggest that you need to train and then you need to be boosted. So we have been applying this strategy to older people with a variety of different conditions. What we have chosen to do is to develop simulations that are realistic simulations of things that you do on the computer anyway.

What is an ATM besides a touchscreen computer? What is a ticket kiosk? Internet banking is obviously done on a computer, and so is internet prescription refill and shopping. So, because we're interested in severe mental illness as well as healthy aging, we didn't want to have to rely on telling people we want you to pretend that you're at your ATM. We put them at their ATM, we tell them what their PIN is, and we say, enter your PIN and check your balance.

The training -- the software was originally developed with funding targeting schizophrenia. We published a couple of validation studies, including one showing that there was sensitivity to impairment on the functional skills task, and correlation with other common outcomes. We recently published a study showing that there was better functional capacity and cognitive performance in clozapine responders compared to non-responders, in a cross-sectional study.

But our primary funding recently has been from the National Institute of Aging. What we did in this study was we did a randomized trial which started out with 140 older individuals, half had mild cognitive impairment, the other half had normal cognition. It was a randomized trial where people received skills training alone or skills training combined with computerized cognitive training using the BrainHQ Double Decision training algorithm, which is exactly what they used in the ACTIVE trial. There were up to 24 hours of total training. When you mastered the simulations, we had you stop. We didn't have people train past the point where they had already mastered what was going on.

There's a lot of details to discuss and not enough time to do that here. But as we see, there was clear sensitivity of these simulations to the cognitive impairment seen in MCI individuals. The impairment was between 1 and close to 2 standard deviations, compared to healthy individuals tested on the same test. So we're seeing that the impairment is substantial. In fact, interestingly enough, in this paper, we showed that impairment on a neuropsych assessment was essentially equivalent in magnitude to the impairments we're seeing in these functional capacity measures.

We did see, though, that when we trained people for up to 24 hours, the proportionate improvement in their performance on the task was the same, in the participants with normal cognition and impaired cognition at baseline. Obviously the impaired cognition people started out quite a bit worse, but on average, at the end of training, the average level of performance on every stimulation was higher for the cognitively impaired participants than it was for the healthy individuals at baseline. So there was a broad training benefit.

We also found that there was a facilitating effect of cognitive training. People who received computerized cognitive training only got half as much skills training as the people who were randomized to skills alone, but the improvement in the functional skills task with half as much training was the same across every single task, when the combined therapy was offered. So what we're seeing is that augmentation of skills training with cognitive training leads to a substantial benefit, and conversely, we see the same thing with cognitive outcomes. There's a synergistic effect on the cognitive outcomes associated with being randomized to both skills training and cognitive training, keeping in mind that the skills and cognitive training is each delivered for only half as much, in terms of the number of sessions, as the skills training alone was delivered.

And what we're seeing is very nice improvements in cognitive functioning. They are facilitated by the combined therapy. And it really is encouraging to see that in a population who commonly couldn't remember the last time they'd been in for training, that we're getting these gains. They were persistent at a follow-up. We were interrupted by the COVID crisis, but we have found 60 of our participants, out of the 96 that completed, and they have agreed to come back and be reassessed what would now be two years later. So stay tuned, we'll see how persistent this benefit is over a two-year follow-up period.

Another new development is pharmacologically augmented cognitive training. We saw TMS-augmented cognitive training. We saw cognitive training augmented by functional skills training. But a number of studies have paired pharmacological and CCT interventions. They have bene done by the pharma industry, pairing a Gly-T1 inhibitor with computerized cognitive training. They have been paired with memantine, looking at cognitive training with memantine versus placebo in people with schizophrenia.

Guanfacine augmentation of computerized cognitive training was found to lead to a significant improvement in participants with schizotypal personality disorder compared to placebo augmentation. And vortioxetine was found also to improve the cognitive benefits of computerized cognitive training in a randomized trial, looking at older individuals. These were people who had age-related cognitive decline, but not depression. And they were treated with vortioxetine or placebo, versus cognitive training, and there was a statistically significant separation.

So I'm going to close by saying that I think that these combined interventions are clearly going to be important across multiple different domains and populations, doing something in addition to cognitive training, whether it's skills training, whether it’s stimulatory interventions, whether it's pharmacological augmentation, these things have all shown to add to the benefit. And further, given NIMH's interest in identifying treatment targets, pharmacological interventions that have a clear target are easier to interpret in terms of add-on augmentation than asking people to train in two different things, where neither one of them may have a clear neural signature that you can identify in this population.

I'm going to conclude now. I'm going to pass along the baton to Carol Tamminga, who's going to lead the discussion across the different presenters.

CAROL TAMMINGA: Thanks a lot, Phil. Just like some of our previous people, I'm going to ask each one of you a question first while I'm looking at the Q&A. I'm going to start from a backwards direction, Phil, so I'll start with you since you're right here.

Phil, all of your data are very interesting, that skills training boosts cognitive remediation. You said that in many different ways, and you showed many different datasets in many different populations. At first I was thinking the mechanism was just double training, but you said if you hold constant the training time, you still get the augmentation.

DR. HARVEY: It is pretty widely believed that the BrainHQ training may induce neuroplastic responses. That's the selling point of BrainHQ, is that it's a neuroplasticity-focused intervention. So as a consequence, I think that there's a possibility that's what's happening.

The other things to keep in mind, though, is that pairing the skills and cognitive training and looking at cognitive outcomes is actually based on studies of healthy older people who are being trained in cognitively demanding skills tasks, like digital photography or even knitting led to improvements in cognitive performance, specifically in the area of executive functioning. So I think there is a domain-specific benefit from the right kind of cognitive exercise.

CAROL TAMMINGA: Phil, would you say that working in science, like all of us do, that are presenting and that are here, is a kind of anti-aging activity?

PHIL HARVEY: I certainly think so. There doesn't seem to be much of a rush to retire on the part of the people I went to school with or even who taught me in graduate school. Everyone who was a professor of mine at Stony Brook who's still alive is still going to work.

CAROL TAMMINGA: Thanks, Phil. I am going to turn to Dr. Rajji now. Dr. Rajji, in your thinking about neuromodulation, you did talk about TMS, and you talked about TDCS. I have a question about intracerebral stimulation, and wonder if you were to do an intercranial stimulation to modulate this beta-gamma coupling that you talk about in improved cognition -- which area of the brain, if you were advising a neurosurgeon, how would you do that? What would be your advice? Which area would you stimulate or inhibit?

TAREK RAJJI: That is a great question. Part of this answer, I will tell you, is actually informed by some of the work done by Dr. Howard Eichenbaum and Robert Greene, who was my mentor when I was at UT Southwestern, doing some training. Obviously, with the TMS and tDCS and tACS, we target surface cortical areas, which is mainly -- we're interested in the prefrontal cortex and that's been a common target. But we know that the networks, especially the prefrontal hippocampal network, is quite important in promoting working memory, but also supporting these cross-frequency coupling and oscillations, and across the network long-range low-frequency oscillation, like in the theta range. And some of the work has been shown by Dr. Gorden and his team about how the oscillations in the hippocampus, preclinical models, oscillations in the hippocampus, how that related to oscillations in the prefrontal cortex, and how the coupling could be also across regions. So what I talked about is mostly local coupling at the site of where we're stimulating.

So we can talk to a neurosurgeon and try maybe to stimulate there, but also some of the work we'll be starting soon is combining electrical stimulation with focused ultrasound, which can -- we don't need this neurosurgeon to target deep. It has the advantage of targeting focally. It would be one of the early work to see if we can stimulate the hippocampus bilaterally with focused ultrasound, and see if there is synergistic effect by targeting the two ends of that network.

CAROL TAMMINGA: Very interesting. Thanks a lot. I will come back to you later on.

Dr. Bhattacharyya, of course you talked about cannabidiol, which is interesting to all of us. One of the questions that I have about CBD is whether CBD is a kind of a treatment that might not be effective in everybody that has psychosis, but maybe in a subgroup. How have you guys thought about this, and what is your own thinking about this?

SAGNIK BHATTACHARYYA: Thank you. Again, a very important question. I will first give a short answer. The short answer is we don't know yet. It's very possible that it may not work in everyone. And the reason that we don't know this is because there haven't been large enough studies to look at it closely, those who respond versus those who don't respond. But hopefully we have now a number of studies, large enough trials, that might help answer those questions.

CAROL TAMMINGA: That's very nice, thank you. I'll be back to you with more questions after a while.

On to Anissa, who was out first speaker. Anissa, I have kind of a broader question for you than exactly a question about the opioid receptor or the D1 receptor. Do you think that psychosis, in terms of all of the patient volunteers that you used, do you think that psychosis has a common pathophysiology? Or do you think that there are multiple pathophysiologies? So that could you find a kappa opioid-sensitive person with psychosis, and at the same time find another person who's sensitive to D1?

ANISSA ABI-DARGHAM: Yes. I mean, absolutely. I think that what you're raising is probably true, that there are different etiologies to psychosis, different etiologies to any of the cognitive, any of the symptom domains we see in schizophrenia. The big question is what are they and how can we define those subgroups?

I think that in the absence of knowing that, we are not likely to succeed very well. I think we were lucky with the D2 hammer type of approach, that somehow dopamine dysregulation ends up being maybe present in a big majority of people, and the dysregulations that they have, whatever those are, lead maybe to a common -- it's almost like the excess dopamine in the striatum could be like a fever for many of those subgroups, and we were lucky it was the D2 blockers there.

But you know, like you don't treat hypertension with just one medication. If you treat it with one medication, it's because you are targeting the very end endpoint. But if you really want to be mechanistic, you just kind of have to know how do people develop their specific type of hypertension, and we're not there with either psychosis or cognition or negative symptoms.

I was thinking the same when you were just asking Rajji where would you target the stimulation. I was thinking it really kind of is going to depend what is it you're trying to achieve. As we heard earlier, it's not only a global brain disease, it's possibly a whole person disease. But let's consider it's a whole brain disease, we have so much evidence for that. We're going to need to develop multiple approaches to multiple types of symptom domains for multiple etiologies. It's a big task.

CAROL TAMMINGA: Anissa, sometimes people are now targeting sort of with neuromodulation a circuit, just like Dr. Rajji just suggested. Could you think of doing the same with drugs, I think? That you wouldn't necessarily be targeting a receptor with a drug, as much as a circuit. I don't know how you would do that, really.

ANISSA ABI-DARGHAM: Yes, you could do it kind of serendipitously. Like let's say this kappa antagonist actually materialized and we can test it. We will find out whether it works on the anhedonia in schizophrenia, as it worked in the anhedonia of anxiety, and it also somehow works on the psychosis. And I think we have to work it backwards. If we are lucky enough to be able to combine imaging biomarkers with these kind of therapeutic approaches and see what changed where and corresponds to which improvement, that would be very fortunate.

CAROL TAMMINGA: Very interesting. Since we're right now at the next question and answer point, I'm going to invite the people on the panels to come up with questions for each other, and invite in the additional chairs, like Dr. Frangou and Dr. Lee.

Does anybody on the panel have questions for another one of the panelists, that you'd like to ask now?

TAREK RAJJI: I have a question for Dr. Harvey actually about the very exciting work of combining skills training with cognitive training, and I wondered if you're seeing any moderating effects that are specific to this combination, versus during each intervention alone? At the same, are there some mediators that are specific to the combination or synergistic effect of the two treatments?

PHIL HARVEY: We were expecting to see much more of an interaction between baseline cognitive status and gains, and we didn't see that. I think part of that was because we picked the absolute best cognitive training algorithm that had been shown to work so well previously, and I think that another thing that's really important is what our result suggests is that the human factors engineers who are building these technological interfaces really know what they're doing. It's actually pretty easy to teach people to use an ATM or to do online banking, even if they’ve never done it before, because a lot of it is either cognitive impairment or it's lack of previous exposure.

So we had a lot of people who had never done internet banking, even people with a lot of money, but they were able to learn it fairly fast because the interfaces themselves are intuitive. That was one of the reasons we picked them, was to avoid having to get people to pretend. Because, with serious mental illness, you ask someone to pretend, you never know what they're pretending, right?

CAROL TAMMINGA: I am going to extend that question a minute, Phil. Is there a pace of cognitive remediation and skills training? Do you have to give people some space in between intensive training period to let their brain activity wander, like people sometimes talk about that. Like sleep, and quiet reflection, or something like that.

PHIL HARVEY: The results of meta-analyses suggest that's absolutely critical, that you get fewer training gains if you train every day than if you train three days a week. And that's applicable to healthy older people. So what we tell people, we're doing a follow-up study. Our software is not completely deliverable remotely. So we've sent people home with the devices, and we're having older people with MCI and normal cognition train at home, and what we did is we programmed the system so they couldn't train more than three days a week. So if they try to log on Monday and Tuesday, it tells you to come back Wednesday. We use the technology to do that.

Plus, the pacing is critical, because the training is paced. You operate at your own speed, and you speed up as you get better.

CAROL TAMMINGA: I see. So your background is just on your off-training days -- you're not down by the pool all the time.

PHIL HARVEY: No, it is true. But what we do tell people to do, and in our new study, we're using ecological momentary assessment to see how much they're actually doing, the stuff that they're trained on in between. So we tell them, try the stuff out, and every day we page them and say did you go on the internet? Did you use your mobile phone? Did you refill a prescription or do any banking?

ANISSA ABI-DARGHAM: Carol, I have a question for Dr. Bhattacharyya. There are some quite convergent PET studies showing that the CB1 receptor is potentially lower. There is just one study showing it's higher, but more so that there's kind of consistency, I think, that it is low. Would that be relevant to your approach with the cannabidiol treatments? Is this something you have to think about? And how would you think about it? What does the drug exactly do? What's the mechanism? Do we understand this?

SAGNIK BHATTACHARYYA: Thank you. Very important question. But again, perhaps we know very little at this stage. So we don't know how CBD, what are the molecular mechanisms underlying its potential, I think I would say, antipsychotic effect. So there are a number of possibilities, so you mentioned about CB1, and again, with CB1, CBD, we know is a functional antagonist, so very likely kind of an allosteric inhibitor. So, a negative allosteric modulator of CB1 receptors. There are other potential targets. For example, CBD might inhibit the hydrolysis by having an effect on FAAH, the enzyme that is involved in hydrolysis of anandamide, which is one of the main ligands, endogenous ligands for CB1 receptor. So again we don't know whether that is the mechanism. Again, Marcus Leweke's study showed that perhaps that is, though we have not yet published the data. We have not been able to replicate the effect of CBD on anandamide levels in the study that I briefly showed you.

Then there are other potential targets, like, say, 5-HT1A, which have also been touted. So it's an important question, but we don't really have the answers yet.

SOPHIA FRANGOU: There was a question in Q&A from Dr. Gault from our audience, that it's relevant to what we're discussing, and I think we can talk to it now. The question is, what is the role of CBD in people with schizophrenia, schizoaffective disorder, that are regular cannabis users?

SAGNIK BHATTACHARYYA: Again, another very interesting question. Because at least in South London where I work clinically, a fair proportion of my patients with psychosis use cannabis fairly regularly. They often try to convert me into using cannabis myself. It's a very pertinent issue. I think if you go by the healthy volunteer studies where we gave people THC and then used CBD pretreatment, we would expect CBD to mitigate some of the harmful effects of cannabis use. We have not formally tested this. We have also done a small study in people with established psychosis where we gave them CBD. I showed some of those results. Some of the participants in that study were also cannabis users, and we saw that there was some effect on symptoms, which I didn't present today.

It's too small a sample to separate out those who didn't have comorbid cannabis use and those who had comorbid cannabis use, to be able to say definitively. But definitely worth pursuing that line, using it as a treatment to mitigate the harm from comorbid cannabis use.

CAROL TAMMINGA: Dr. Frangou, do you have any questions of your own?

SOPHIA FRANGOU: A couple, as per usual. One is to Phil Harvey, and it's really not about so much the science, but the practicalities of delivering this intervention. If you see deployed in the context of the routine clinical service, are the requirements in terms of personnel too high to actually become routine? Or is it easy to incorporate it into a routine setting? Or maybe you've done that already, I don't know.

PHIL HARVEY: Our whole goal is to roll it out and have it be broadly accessible. The current version of it is completely cloud-based. So what you can do is you can teach people in a group how to train themselves, on both BrainHQ and the skills training software and send them home. And then the login is a single login for everyone in the whole study. They just go to the website of the startup, and login, and then they can only follow the path that they're allowed to follow.

If you're training, when you log in, it takes right to the last place you trained. We developed the software such that when you'd mastered something, you didn't train on it any more. If you were stuck on something, it pushed you forward, then brought you back. And every time you come back to retrain after the last time you trained, you start on the last thing you got right. So you log back in, and you pass again, and then you move forward.

The pragmatics of it have worked out okay. We have had as good adherence from home as we ever had in the office, because people don't have to get in. Some people don't have devices, but we found a place to buy Chromebooks for $75 and loan them to people and give them a cellular hotspot to take home with them.

SOPHIA FRANGOU: I will follow up with a question that doesn’t come from me directly, but I'd like to pose this from Dr. Kotov to you. And it says, is it correct to conclude that speed training improves cognition generally, and other types of cognitive training, for example, memory, reasoning, and so on, we can do without?

PHIL HARVEY: Well, I think that the results of the ACTIVE trial did suggest specificity of benefit. But I don't think that the speed training made you better necessarily at problem-solving. But the problem-solving training didn't persist, and the problem-solving training wasn't correlated with ADL gains that were persistent at a decade out. So if you're going to train something that has the biggest impact and lasts for the longest time, that would be speed training. But we don't know, and in fact, we did speed training and skills training, and the biggest improvement was in episodic memory in people with amnestic MCI. So I think there's a message there, but I think that's a point that's not yet proven.

But I tell you, though, the speed training is also the stuff that's easiest to get people to do, because there's a second algorithm called Hawk Eye, which is like whack-a-mole, when you go to the carnival, and that actually was recently used in a successful study of treating traumatic brain injury. So it is the stuff people like the most. They don't like memorizing shopping lists and stuff like that.

CAROL TAMMINGA: Dr. Lee, do you have any questions?

ELLEN LEE: Yes. This is a little bit more broad for all of the panelists, all the speakers. I was just wondering about the ancillary effects of many of your treatments, many of them would facilitate potentially human interaction, pain modulation, sleep-wake cycles, other effects that could also help with symptomatology and functioning, and I was wondering what you thought about the other side effects of your medications and treatments.

TAREK RAJJI: Maybe I can chime in. For the brain stimulation studies where we did that's a great question, because we do monitor quite a bit the effects with respect to headaches, with respect to balance, with respect to sleep. And some side effects are known, that some people would complain temporarily to changes, let's say, in their sleep or changes in headaches.

So what would be I think interesting to look at is if we start again combining more of these type of interventions with quite diverse type of treatment for the whole body, and not just kind of thinking -- so we've been doing a lot of the work combining brain stimulation with cognitive training, with other forms of brain focus intervention, but maybe we need to think beyond that, so kind of agreeing with the previous comments about that these disorders are not just brain disorders.

PHIL HARVEY: In the ACTIVE trial, they discovered that people got better at driving. Because, actually, there's a history there. The double-decision task is actually a neuropsych test that's called useful field of view, which has been adopted by motor vehicle departments in over thirty states in the United States. Because as you get older, your peripheral vision isn't as good. So having impairments in your peripheral vision is associated with crashing. So in the ACTIVE trial, they discovered that people who did well with the speed training drove better, had fewer accidents, and were less likely to stop driving. However, unfortunately, no one started driving for the first time at age 75 after getting speed training.

CAROL TAMMINGA: Well, with this I would like to thank all the participants in this last session. I thought it was just a marvelous session and I learned a lot. I learned a lot in all the sessions, but I particularly am grateful for this session, and with this I'd like to turn the meeting back over to Dr. Frangou and Dr. Lee, and it says here that you guys are going to synthesize everything. You're going to identify the opportunities, and you're going to tell us what the next steps are. I think we ought to give you plenty of time for this.

ELLEN LEE: Many thanks to the speakers today. Just to thank everyone who spoke, and we want to provide a little bit of a wrap-up and sort of a synthesis of what we discussed today. Our session one from yesterday, we were talking about phenomenology, course, and outcomes. One of the common challenges that were reported were how do we best define late onset schizophrenia? And really the discussion that came from this was that the current approaches really use variable age cut-offs that don't necessarily have a robust, empirical foundation. We really need these data-driven approaches based on large registry data that is harmonized across multiple populations to really capture the heterogeneity of symptomatology and presentation.

Another challenge was does the onset of schizophrenia in early, versus mid to later life, define subgroups with a distinct etiology or pathophysiology? And it really does appear that there's evidence that later onset schizophrenia is associated with less social and occupational disability and a lower genetic or familial risk.

Another challenge was do clinical symptoms of schizophrenia, positive or negative, change as individuals with early-life onset reach middle and late adulthood? And there is great evidence that suggests significant interindividual, interstudy variability. We learned a lot about the different types of definitions of remission and recovery and how different assessments can be more sensitive or specific to some of the deficits we're seeing.

I think, again, this is just another argument to really combine forces and really understand how this heterogeneity and the difference in presentation can be helpful for us in clinical practice.

Another major challenge was whether the treatment needs of women with schizophrenia were being met by current practices, and Dr. Sommer's talk really showed evidence that the first-episode services are not necessarily meeting the needs of women, in terms of timely detection and supportive interventions that would be more focused on, for example, families and relationships that are more pertinent to them. As well as the fact that antipsychotic medication doses doesn't necessarily account for the differences in pharmacokinetic and pharmacodynamic principles. Women may be more sensitive to medications or metabolize them differently.

Other major challenge that was brought up during this session was what is the role of estrogen and other analog treatments in schizophrenia? While there is some early evidence for beneficial effect in women, particularly for cognition, I think it's really promising to think about how to extend these findings to larger populations and to focus these treatments and make them more personalized.

Moving on to session two, there was a great session on cognition, including all domains of cognition as well as social cognition, and some of the key challenges brought up there were is there a higher risk for dementia in schizophrenia? And longitudinal studies really suggested a gradual decline that may accelerate with aging, and that there seems to be a contribution of antipsychotics or duration of illness. However, many of the panelists and speakers brought up the key methodological challenges to really understanding this cognitive decline and to really assess more robustly for neurodegeneration in this population, both by having older adults and following them for longer periods of time.

Another challenge was how to best disentangle the effect of illness versus treatment on cognition and schizophrenia, and how to best address the issues of sample diversity. And Dr. Yang spoke about his study in China, which looked at treatment-naive schizophrenia patients, which really suggested a negative effect of longer illness duration on cognition, especially in these rural, undereducated populations. And this really highlighted the need for more studies in low- to mid-income countries to really capture that heterogeneity of aging in different environments.

Another challenge was whether different cognitive domains have distinct trajectories in schizophrenia as they age. And the evidence here was suggesting that decline may be more pronounced in neurocognition, compared to emotional and social cognition. There were key relevant cultural differences that were raised by the moderator and the speakers, which I thought was really important to think about as we expand our types of studies that we do.

Another challenge was are there changes in the brain that are associated with cognitive function in schizophrenia, and do these differ as patients age? There is some evidence here for differential involvement of the hippocampus during different stages of the illness, showing anterior-posterior decline in volume and transitions from hyper- to hypofunctioning connectivity, and the association of these changes with aging or other processes are really important to consider.

Moving on to session three, where we talked about regulatory metabolic processes that might underlie aging in this group. There was one of the challenges was talking about how sleep spindle density was linked to thalamocortical connectivity and memory consolidations, as well as how spindle density declines with age. And the current data suggested that reduction in spindle density was linked to abnormal thalamocortical dysconnectivity in schizophrenia and that increasing coordination of spindle frequency and other oscillatory network activity could really provide a novel therapeutic treatment target.

Another challenge was how poor sleep quality and, in particular, obstructive sleep apnea, is an underrecognized condition in schizophrenia, and is associated with medication dosing. And the evidence there was that OSA may contribute to clinical and cognitive deteriorations via multiple biological pathways, including oxidative stress and inflammation.

Another focus of this panel was about the challenges of how energy production and regulatory pathways were implicated in schizophrenia across the lifespan, and these speakers spoke about how transcriptomic analyses suggest that lower ATP in schizophrenia and pointed to potential treatment implications of targeting such pathways.

Also, imaging data suggested reduced ability in patients to generate ATP, the shift to glycolysis, the lactic acid buildup, a breakdown in the ability of ATP to sustain these long-range functional connectivity in schizophrenia, as well as bipolar disorder, as well as the involvement of these pathways that can change over the course of schizophrenia. And there was some exciting evidence for NAD-related treatments as well.

Switching it over to Dr. Frangou.

SOPHIA FRANGOU: Here we focus on the sessions that we had today on accelerated aging, and the challenges that we identified, I think could be summarized as what is the evidence for accelerated biological aging in the periphery and in the brain themselves? Are they linked? The presenters gave us a wide range of markers that have been associated both with disease and aging, and in some cases with their interaction, that implicate multiple pathways across body and brain and sort of underscored the complexity of the landscape that we have to work within.

The second thing was focused a bit more how we measure aging, accelerated, advanced, or whatever, using brain imaging, and we had the benefit of the efforts of the ENIGMA consortium, that has been able to amass a huge amount of data comparing healthy individuals and patients with schizophrenia in terms of brain age, which is a very interesting measure because it's also very personalized, talking about sort of precision medicine. Each individual has its own specific point of deviation from the typical trajectory and generally reinforces the point that there is evidence for brain accelerated aging in patients with schizophrenia.

And then when we go to Hilleke's talk, this general picture was complemented by the fact that this aging process doesn't seem to be uniform. It seems to be a little bit more accelerated early in the disease stages, immediately or perhaps within the first five years post-onset. It is entirely possible that there are secondary periods of increased acceleration, when people get even older, but we don't have data going that far. But importantly, interventions that have been known to improve brain health, let's say, generally, in the population, like exercise, seem to also be relevant in patients with schizophrenia and potentially an avenue of intervention, either with specific interventions or thinking about how we structure environment within cities and neighborhoods to enable such activities.

And the final talk, the final challenge and also the opportunity, was associated with the development of new methodologies like EVs that were presented by Dr. Kapogiannis. These extracellular vesicles have been validated, to my mind, at least, in terms of their usefulness in capturing many active changes in molecular pathways within the brain while people are still alive. And therefore we can actually look at dynamic changes associated with this progression; we can also look at dynamic changes associated with treatment response, and even at some point trying to stratify patients based on what sort of pathway may be more or less implicated in the pathophysiology of psychosis in their particular case and try to match that with some aspect of treatment.

In session five, we focused even more on established and emerging treatments in schizophrenia, and I think there was a general agreement that not all of these opportunities for improving cognition and clinical severity in schizophrenia have been applied to populations with later onset or to patient populations with earlier onset but who are sort of further down in terms of their disease trajectory.

We saw a range of opportunities to intervene at different systems, the dopaminergic system, the opioid system, the cannabinoid system, in terms of pharmacological interventions. We also saw the value perhaps of neuromodulation and thinking about how we can intervene in more targeted and more informed way to change the functional dynamics within the brain to improve cognition -- across the board, but also in older individuals. And of course, how cognitive remediation has really advanced quite a lot in terms of its specificity and optimization, both in terms of the skills and targets of the training, but also in terms of the ease of its delivery to a wider patient population as possible.

And very encouragingly, we actually saw that -- and that's the other challenge -- we saw that very encouragingly, these changes can be beneficial for people across the entire age range, which suggests that they will likely be very beneficial for older patients with schizophrenia.

A specific consideration, of course, was given to the issues of tolerability as people get older. Patients are not -- we all accumulate allosteric loads, we accumulate a number of other problems that may interact with the disease trajectory for schizophrenia and it may impair function, but also our ability to intervene successfully.

I think everybody has said throughout the entire workshop that there was a wealth of information that is already being accumulated, suggesting the need to address both the clinical and the research needs for people that have a later onset of schizophrenia and for people with schizophrenia as they grow older in general.

I think the biggest need that was identified, given the complexity of the systems that we have seen, has been in ensuring that we optimize what we currently have through perhaps consortia, networks, and harmonization of existing datasets, to increase our return. But also to think of working together to acquire largescale longitudinal data on these populations.

There was an emphasis on standardization of instruments, that a lot of the interstudy variability has to do with how we conceptualize either phenotypic definitions or the different outcome domains. And of course, a very important focus was on sex, and particularly female sex, given the fact that there seems to be a secondary peak later in life that is very specific to women that is not sufficiently recognized either in terms of early detection, but also in terms of the special needs that individuals, females, may have specifically.

Again, just to recapitulate the core messages: optimization, collaboration, consistency of assessment, quick translation of research into novel treatment development, are the very broad areas where the challenges may meet opportunities and where we think this particular workshop has helped people think about how to position themselves, think about collaborations, and think of different or more innovative study designs.

On behalf of Ellen and myself, and of course Laura, Andrea, and Craig from the NIH and the entire -- that supported this workshop, I want to thank you for your very enthusiastic participation to this workshop, and I hope you found it enjoyable and informative.