Welcome to our session, Clinical and Neurobiological Predictors of Substance Use Disorders in Youth with Disruptive Behavior Disorders. I'm Jeffrey Newkorn, and I'll be chairing the session. And I have three outstanding presenters in this session, Ilyan Ivanov, Brooke Molina, and James Blair. Each of them has a very robust interest in the area that we're going to be talking about today. Ilyan is, I'll get out from under this thing. Nobody will ever see me. Ilyan is a professor of psychiatry. He's the site division chief at Mount Sinai Morningside. He's a friend and a close colleague of mine for 20 years. Brooke Molina is a professor of psychiatry, psychology, and pediatrics, director of the Youth and Family Research Program at the University of Pittsburgh. She is the president-elect of APSARD, which is the American Professional Society of ADHD and Related Disorders. And you'll see at the end of her talk what that is. And she's also been a colleague of mine for quite a number of years through the MTA study. And James Blair, visiting us from Copenhagen, is professor of translational psychiatry and leader of the biomarker group at the University of Copenhagen. He's really a fantastic expert in the area of conduct problems and biological mechanisms underlying it and risk for related disorders, like substance use disorders. He's been a collaborator, too. So it's a fantastic group. And I'm going to turn it over to Ilyan right now. Thank you all for coming. As Jeff mentioned, I think it's a very exciting session. In terms of discussing risk factors, and what I'm going to talk to you about is kind of a continuation of a session that we had in 2019, that goes all the way before the pandemic, when I spoke in that meeting about the background for the research that … When I was discussing the background for the research that I'm going to present today, so there is a chance that any of you at that session, some of the background slides might be familiar. Okay. These are the disclosures, no industry support, and in terms of the objectives, I just want to, as you're reading through that, emphasize something. Not at any point we're suggesting with the research that we engage into that stimulants may be paving the way for substance use in patients with ADHD. The stipulation is that within the larger cohort of ADHD patients, there might be subgroups with different neurobiological profiles, and what we don't know is how stimulants, who might be introduced as treatment early in childhood, might interact with this predisposition. So the studies that we conducted is actually looking at that, just keep in mind. So just kind of as a form of introduction, but any known substance of abuse, in addition to behaviors that associate it with like pleasure and some hedonistic component in it, associated with increase in the dopamine in number of brain neurocircuits, especially what's called motivational reward brain circuit, and in that account, stimulants are in the same category. Now there is this idea of the reward deficiency syndrome, stipulating that low dopaminergic states may present a vulnerability, meaning a hyperactive brain system might be hijacked by substances of abuse that have shown to increase the dopamine release about tenfold over natural reinforcers, like a favorite food or some pleasureful activity. And the reward deficiency syndrome is this umbrella construct that some people think is associated with genetic predisposition and is a risk factor for later substance use. Conditions like ADHD, conduct disorder, and substance use, the best we know about their neurobiology actually fit that category. This is a model that was developed in 2014. It's a meta-analysis of the data that was available at the time from neurobiological research in ADHD specifically, and the authors, you know, rightfully so concluded that ADHD is indeed associated with hypoactivation in the brain reward circuitry, although these are not studies that are particularly focused in reward processing, but there is an overlap between areas that have been influenced by the stimulants, and these are also areas that are associated with reward processing. And what you can see here, I'll use my cursor if you can see, no you can't, so in the panel A, this is this inverted U-shaped curve suggesting that the relationship between impulsivity measures in the X-axis and activation in the ventral striatum in the Y-axis has this linear positive relationship, meaning if we're within the normal population, normative population, not diagnosed with ADHD, and if you have high impulsivity scores, you know, individuals on the spectrum, that is associated with elevated activation in the ventral striatum, however, individuals who pass the peak to the right side, and now they have symptoms that qualify as diagnosis of ADHD, the relationship becomes negative, more impulsivity, less activation, and as you can see to the right, the two groups of individuals kind of nicely fit into two separate clusters. So what is the possible link between stimulant treatment and SUD? Now ADHD by itself, although not considered like the highest possible risk, but it carries some independent risk for substance use disorders in life, that's related to both disrupted behavioral control, reduced reward responsiveness, and following that model, actually stimulants who might be restoring the level of dopamine release in the system, in the brain neuropsychiatry, should be helping those individuals. Now ADHD often exists in comorbidity with conduct disorder, and conduct disorder by itself is actually quite, you know, carries much higher risk for subsequent substance use. So the studies that look into reward responsiveness in ADHD with conduct disorder, just comorbid conditions, been kind of inconclusive. There's actually a number of studies that have come out showing that some individuals may show low activation, some individuals may show high activations, and that depends on different reward conditions. So it's not like one ubiquitous type of response in ADHD, it varies among different individuals. So, it's actually possible from theoretical standpoint that stimulants may affect subgroup of individuals who may have somewhat kind of a different neurobiological makeup. And that phenomenon is called sensitization, it's been studied extensively in animals, not necessarily related to substance use, it's more related to designing different animal models, for instance models of schizophrenia, and they're using measures that are not relevant to substance use, they're relevant to things like locomotion. So it's not a direct linkage, however, in general what sensitization implies is that if you expose a system, biological system, to a certain agent at point A, and re-expose them at time point B, if the system becomes sensitive, it would have a higher behavioral response, or you can elicit similar behavioral response with a less amount of medication. So that's the sensitization paradigm. Our group actually has, including James and Jeff here, have published a couple of papers that outline this theoretical model. One is in the Neuroscience and Behavioral Reviews, talking about a very focused review of ADHD literature. I won't be reviewing that, but if you're interested. And one of our residents also looked into the possible mechanisms and the interface between amphetamines and the opioid system. And I can summarize in this citation what we concluded in these papers, but in general, this comes from a 2019 paper in Biological Psychiatry, that is a meta-analysis on ADHD and stimulant treatment. And it's a nice summary of the state of research at the time, which is plenty of animal research studies have suggested that sensitization to stimulants is an actual phenomenon in animal models. However, clinical studies, like naturalistic follow-up studies, like MTA and many others, have shown that there is a neutral effect, meaning ADHD individuals treated with stimulants don't turn out to have higher rates of substance use. Even more so, more recent studies from epidemiological research shows that treatment of stimulants in adolescents and adults would be beneficial in respect to outcomes that are related to substance use, like visits to the emergency room and accidents and things like that. So clearly there's a disconnect between the human and the animal research. Moreover, the sensitization phenomenon in humans actually haven't been studied. One of the obstacles in that is that sensitization markers are very difficult to outline. So Robinson and Berridge published a series of reviews stipulating that unless you're able to index neuroadaptation at brain level, sensitization would be very challenging to establish otherwise. So you need kind of a biological marker. As the behavioral markers are kind of notoriously unreliable, especially in younger kids. Based on that, there is this paper that came in 2006. It's a PET study in 10 volunteers using a sensitization protocol, and they use metamphetamines in three subsequent scans. And the second and the third scan that were concluded after the completion of the sensitization protocol show that there is a persistent neuroadaptation change in the brain reward system a few weeks after the administration of the medication, and the third scan was a year later. So the indication was that there is a sustained change on brain level, supporting this idea that you have to have some way of examining the brain in order to ascertain sensitization. So several reports have come out showing that within the ADHD, large ADHD cohort, if you define groups at different levels of risk, like ADHD with some other clinical risk factor, those may be showing different patterns of brain activation. So this is one paper that we published 2019, and what you can see here is in our high-risk ADHD group, ADHD group family history of substance use versus controls versus low-risk ADHD only, you can see that the green bars is significantly different in these three different brain regions in what is called unexpected non-reward outcome. In short, this is a case when participants expected to win something, but they did not, and that caused some level of frustration. So clearly the high-risk group is different than other ADHD participants and controls as well. And that has been documented by others. This is just to illustrate in this first study, controls and disruptive behavior disorder individuals. That probably includes ADHD and conduct disorder ODD. As you can see in the punishing feedback, the pattern of activation is just the opposite of what is in the reward feedback, which is the upper graph. And something similar in a separate study comparing controls, ADHD, and individuals with, let's say, psychopathic traits here, conduct disorder, I would imagine. And the reversal error punishing condition also showed increased activation compared to this other group. So clearly ADHD is not just one group of individuals presenting with the same level or the same type of brain activation. Based on that, in the paper that I showed you initially, the review paper, we kind of outlined this model suggesting that if you're an individual, somebody who has ADHD and have hyporesponsive brain neurocircuitry, the treatment of stimulants may optimize or correct or compensate for that and potentially decrease risk. No treatment, we know that ADHD individuals have somewhat elevated risk of substance use. If you have somebody or a patient who have ADHD and some cumulative, some additional clinical factors, either conduct disorder, family history, the treatment, very possible, is to kind of exacerbate that hyper-responsiveness or magnify it somehow. That is the premise of the sensitization model, and that was the premise for the study that I'm going to present to you right now. So it's a pilot study. The main objective was to compare neurophysiological response, fMRI response, during a reward task. And in addition to that, we introduced a number of behavioral measures like ADHD symptoms, as well as impulsivity symptoms from different scales that have been related to risk of substance use, like UPPS and Kirby. And the approach was to pretty much run an open-label clinical trial with amphetamine and scan individuals pre and post. So we started right before the pandemic, then the pandemic hit. Finally, we were able to recruit about 18 children. 13 of them provided good data, and these are very young individuals, 7 to 12, but majority of kids are about 7, 8, and 9-year-olds. Our low-risk groups is ADHD only. Our high-risk groups is ADHD with severe ODD or conduct disorder. And these are the measures that we use. It decides to diagnose ADHD and any comorbidities, extensive family history, risk of parental substance use, ADHD arrest, and then Kirby is a delay-discounted scale pre and post. And UPPS is another impulsivity scale that tests or indexes things like urgency and perseverance and sensation-seeking and things like it. So the primary outcome, as I mentioned, was our neuroimaging outcomes, and I'll just very briefly outline the task that we use. It's a variation on something that's called MID, monitor-incentive-delay task. It has just two components, and in order to understand the outcome, I just want to point the relationship between the dollar sign over there, which suggests this is a trial when the child can win $1, and the outcome is they either win $1, they sometimes lose, sometimes they don't win anything. The novelty of the task is that about 50% of the correct responses, which is this thing in the middle, it's a flanker task, like airplanes going different directions, 50% of the correct responses, as they expected to win, were not rewarded. So it creates that frustration, non-surprising, as we call it, surprising non-reward component. So they have reward and then violation of reward expectation. These are two different outcomes that I'll be presenting data on. This is the baseline comparison between the two groups, and as you can see, the blue, the high-risk group, is consistently showing high activation during the reward component to the left, but predominantly in this surprising violation of reward expectation in the rest of it, in the VOPFC, the insula, and the anterior cingulate. So this is a confirmation of this previous study that I showed, separate, it's an independent sample, small sample, still similar pattern of high-risk individuals showing high activation during different reward outcomes. This is what happened after the treatment, and this is in the reward component of the task. So the participants won this $1 that they've been playing for. Here you see that our low-risk group, individuals with ADHD only, have significantly larger change. So this is delta. Delta means activation at the end minus the activation at baseline. So all the positive red bars are showing you the activation at the end of the study for the low-risk group was higher in this network of ventral striatum, ACC, amygdala, and ventral lateral prefrontal cortex, where the high-risk group show either minimal change towards decreased activation or almost no change at all, like into the ACC. When you look into the other outcome, which is the violation of the expected reward, we see just the opposite. Now, the high-risk group, the blue bars, are showing that in the end of treatment, this individual showed much higher activation, significantly higher activation in the similar network, almost identical, amygdala, ACC, ventral lateral prefrontal cortex. As you can see, the low-risk group has somewhat kind of a minimal change. So let me quickly summarize. Within the reward component of the task, when they win positive reinforcement, we have increased activation in the low-risk group, which might be a possible correlation for pre-existing hypoactivation in ADHD. We actually have a study in adults that we published in February showing very similar results. And when it comes to the frustrated non-reward, the violation of the expected win, there was an upregulation in this corticolimbic network with the high-risk groups. And this direction of the slope shows you this group-by-condition interaction. So I loaded your information, and if you're thinking, like, what is the meaning of this many musicians actually have asked that question. So here might be the answer that is helping you to interpret this. Let me present, like, the clinical data. So this is the ADHD pre-to-post. What this table shows you is that all participants from both groups have shown significant reduction, more than 50% reduction, in their ADHD scores. So in short, whatever happened in this VOPFC-ACC-amygdala network did not interfere with the clinical response. Actually, based on clinical response, you cannot differentiate between the groups. What was of interest, though, is these measures of impulsivity that are more specific to substance use risk, like sensation seeking from the UPPS. The high-risk group showed statistically significant reduction. The impulsivity came down. This is even more demonstrative in terms of visual, but statistically not significant. This is the curve, the delay discounting. So we have impulsivity measures from ADHD improving for both groups, impulsivity measures specific to substance use, like sensation seeking from the UPPS, and the high-risk group impulsivity measures specific to substance use, like sensation seeking and delay discounting, improving for the high-risk group, and that's the differential response. Going back to the neurobiological underpinning of all this, if you try to connect this, this is a well-established VOPFC-ACC-amygdala network implicated in emotion regulation and top-down behavioral control. A number of papers written about its relevance to substance use disorders. One potential hypothesis here is that if the direct VOPFC-amygdala network is malfunctioning or not acting sufficiently, you can have an indirect pathway through the ACC, and up regulation in the ACC can optimize the system. Why is that important? Well, we know from adult literature that actually VOPFC could be a target for both indexing and then neuromodulation. This is a paper in adults with cocaine disorder that shows that TMS activation of the POC and neuromodulation of the VOPFC has positive effect in decreasing cravings and curing activity. So let me kind of conclude. The first point I want to make is that behavioral and biomarkers are important, as opposed to what was initially kind of put forward. Obviously, there is a value for both. And the combination of what I showed you as our results seem to suggest that in high-risk individuals, ADHD and conduct disorder, stimulant treatment has kind of distinct effect on VOPFC-ACC-amygdala network that might be contributing to decrease of impulsivity aspects that are related to substance abuse risk. It's kind of a positive effect. Proximal and distant outcomes are different. What we just showed you is like a few weeks of treatment. These are 7- and 8-year-olds. We would not know until probably 7, 8, 10 years later if they would develop substance use disorders. So combining longitudinal studies with experimental design studies is one way to go forward. And this type of research that we're moving forward with application for NARO-1, including things like connectivity analysis, emotion regulation skills, and things like this, may help to understand better what would be the effect of a stimulant medication, what would be the effect of non-stimulant medications that also are proof for ADHD, what might be the effect of neuromodulation kind of procedures like TDCS or FNIRS or things like it, and maybe develop algorithm for sensitization since now witnessing number of substances like ketamine and hallucinogens entering, you know, a rheumatarium, but we know that these things usually slip down to adolescents and teens. These are the individuals who have helped develop this type of research, and thank you for your attention. We have a couple of minutes to do questions. We'll try to take questions specifically about the presentation with each presentation, and then we'll have some time at the end for discussion. Are there any questions about this presentation? We're right on time, perfect. Brooke? All right, can you hear me all right? All right, nice to see everybody. Thanks for coming in. It's a rough time. It's right after lunch, so we'll just give it a go. Fortunately, Ilya uncovered some of the material, so that should be a nice little precursor to my talk. There's a whole cast of characters, including Jeff, who've been involved in this research that I'm going to talk about. This is not an exhaustive list. If you're familiar with the MTA, you know there are many, many, many, and still more joining that group of people working off of that still useful data set. Ilya uncovered this. I'll just hit it briefly, just to make sure it's in your head. But we do know that kids with ADHD have elevated risk. What I want to point out is just having ADHD does not confer guaranteed risk. It's akin to being a child of an alcoholic, or a person with alcohol use disorder. That's better wording. And we know in those circumstances that those offspring have elevated risk, but they are certainly not sentenced 100% to the likelihood of having struggles with alcohol or other substances themselves. But we all want to keep our antenna up. We have studied quite a few pathways now to understand this risk. My own work has looked at a number of different reasons why, psychosocial reasons in particular, children with ADHD have elevated risk. I'm going to talk about a very specific pathway today that has to do with stimulant treatment. Before I do, I'll give you this slide. Feel free to take a picture of it. For those of you who want citations, they're mentioned in the bottom right. I've published, as well as a number of other people, Ilian has published in this area, Beth Gronemann, a lot of people, Tim Willans, Scott Collins, have talked about a range of different factors that potentially help us understand why children with ADHD have elevated risk. As clinicians, we spend a lot of time thinking about the treatments that we provide. Stimulant treatment being right at the top of the list for ADHD is one that always occupies our mind space. I would encourage you to also, though, think about a plethora of additional reasons why kids would be at risk for substance use disorder. Because it has implications for how we manage the condition. It has implications for how we try to be proactive and preventive from the get-go. We also know that stimulants are not 100% well-received by all patients. We also know that they are not the complete be-all, end-all. They don't rid one of that disorder. Therefore, because of all of these factors, there are quite a few different things that we need to think about when we're trying to be proactive and reduce that risk. Ilian talked about the impulsivity and the sensation-seeking. That's smack in the middle of the slide here with regard to temperament and personality. There are many ways to measure that really interesting construct that isn't just one variable. It's a lot of different variables. And then he talked already about differential response to drugs and also the reward system. And this is actually a wording right out of your paper, Ilian, which I really liked, the notion of tepid recruitment of motivational neurocircuitry. I think all of us who work with people with ADHD can understand this challenge, which is the vacillation in motivation. So one can know how to do something. One can want to do something, but that want in the moment to implement the behavior change is very, very challenging. And stimulants can be helpful with that. I want to draw your attention to the second-to-last bullet there. Common behavioral, academic, occupational, and social impairments. These are those impairments that typically draw kids with ADHD or people with ADHD into treatment and are not always fully addressed with stimulants, yet they are risk factors for substance use disorder. So, for example, we have a paper under review right now that shows this, that the conduct difficulties, but even more so in the long run, it's the academic difficulties, the occupational deficits, and social difficulties that contribute to that risk long-term. Stimulant medication is going to hit all of those. And so when you're thinking about trying to address it, you want to think broadly in terms of your clinical practice. We also know that just teaching content doesn't necessarily translate into action. So people with ADHD can know what they need to do. They can know, and we have papers that show this, they can know that alcohol and cannabis have certain types of effects. They know that. Teaching it to them doesn't necessarily translate into action. So those are factors for you all to think about in terms of your treating people with ADHD and trying to think holistically in terms of managing substance use disorder risk. Now, stimulant treatment. Ilyan set the stage very nicely for this. It can be both protective, but can it also potentially be harmful? It's interesting, and I think, agree, important to think about in terms of subgroup vulnerability. This is such a heterogeneous group of individuals. We really, we've been operating at the level of group averages, not necessarily at the level of subgroups. And so I applaud the work that Ilyan and his team are doing. And that QR code is directly to that paper, the 2022 paper that he referenced. I encourage you to take a look at it. There are, this is a great paper that Steve Farone led. It is a review of reviews of all the big review papers in the field of ADHD. And if you look at this paper, how many of you have seen this paper? Okay, good. So feel free to take a picture of that. It is great for clinicians. It is a long list of bullets. And each bullet gives you a summary of a content area in ADHD. So if you want to know what's the latest at the time of this paper, which was in 2021, what the literature has shown us for this aspect of ADHD, this aspect of ADHD, this aspect of ADHD, you can go straight to the bullet. It's a really, really rich paper. And that review suggested that stimulants are protective in terms of substance use disorder. So have that on one side of your brain while you have also the subgroup question that Ilyan's raising, which I think is an important one. It leads us to want to continue to research this question. So we did it with the MTA. The MTA was at the time when it was launched, the largest clinical trial with children that the NIMH had ever undertaken in the early 1990s. And I was super fortunate to land in Pittsburgh right at that time to train as a clinical psychologist to do my internship. And I came to the University of Pittsburgh with a background in substance use risk. It was a perfect time to start to get involved with this study. The study involved 579 children across six to seven sites. There's one site split between New York and Canada. And it was consistent with the times, mostly boys. There were 20% black and 8% Hispanic Latinae in the sample, largely white, and a third one parent. That's just important to know because we're really trying in the field to be more diverse. At the time, this was pretty good. The kids were randomly assigned to one of four different treatment groups. I'm not going to get into those results. But the important thing is that they were treated with evidence-based medication or behavior management protocol, a really involved behavior management protocol at the time, or both, or sent back out to the community. And that lasted for a little over a year. And you can see the timeline here. We followed those kids all the way up to the mean age of 25, which is the last point at the right side of that arrow. And we had really good retention. We had a little over 80% of the participants stayed in the study from the mean age of 8 or 9 all the way to the mean age of 25. That and the multi-site nature of the study is what makes it so powerful. We had repeated measurements all the way through. So I was the one who was, at the time, leading the questions about substance use because I brought that background to the study. And we began looking at this question at the earliest point that we could, which was when the kids were middle school age, 10 to 14 years old. And we found what our hypothesis would have suggested, which was that children with ADHD do start to dabble a little earlier than kids without ADHD. But look at where those percentages are. It goes back to the point that I was making, which is like being a child of a parent who has alcohol use disorder in their history. It's not a guarantee, but the risk is elevated. We did this again up to the mean age of 17. And you can see that the higher bars, which are the children with ADHD histories, are higher than the bars for the kids without ADHD histories. And I want to draw your attention out to the daily smoking at the top right. Look at that difference. Nicotine use, which at the time was cigarettes, is one of the strongest signals in the substance use literature for kids with ADHD. Virtually every single study finds that there is a substantial difference. By adulthood, it's about 40% versus about 20% without ADHD histories are daily users, multiple times a day. It's a very powerful signal in the literature that just does not waffle around. We also found differences for substance use disorder. But you can see by those red lines, the lines are much lower than use. We did it again at mean age 25, which is the last point. Again, we found that there were higher rates of substance use in the individuals who had the ADHD histories than those who didn't. I want to draw your attention to the weekly marijuana and the daily smoking ones. Those are the ones that showed the statistically significant differences. Now, look at the weekly marijuana. For the kids with ADHD histories, about a third reported the weekly marijuana versus 20% without the ADHD histories. So a third versus 20%. Look up in the blue box. Those are the percentages who in Lily Heckman's paper met criteria for cannabis use disorder. The numbers are lower. The main point I wanna make here is not that they have higher risk, which they do, but look at the discrepancy between use and disorder. It's something just to be aware of. Just because somebody doesn't meet diagnostic criteria for disorder doesn't mean that they don't have some elevated risk that's important. And what is it that's getting into the brain? It's the use. It's not the consequences that are leading to the diagnostic categorization of disorder. It's the actual use. So it's something to think about when you're reading papers. Are they focusing on disorder and potentially under-representing the extent of use that's actually happening? This is just, again, really what Iliam was talking about, should stimulants protect or are they harmful? We have two papers where we looked at this issue of stimulant treatment in the literature, or in the MTA data set. The first one, we looked at it in adolescents. We looked at their stimulant treatment out to age 17 and asked three questions, which you can see in the middle of the slide. For those who were treated, what if they were treated early and did that make a difference in the amount of substances that they reported using by adolescents? We found that it did not. And the reason we looked at this is because there's a Manuza et al paper in 2008 that did find that if kids are treated early, which is what most clinicians tend to believe, start early and that'll have an effect. We didn't find that. What if they were treated in the past year? Does that help? We did not find that effect. What about total time, days on stimulants? Lots of people think, well, it's because they stopped. They weren't on long enough to make a difference. We didn't find that either. So that was the first paper where we addressed these questions and did not find an association in adolescents. So the good news is it wasn't harmful. We didn't find an elevated risk in adolescents, but we also didn't find protection. And so, of course, that led to a lot of questions. And one of the concerns was, what about all of those factors that drive individuals to get treatment that could also be driving their substance use, basically confounding variables that you don't have in these analyses that might actually be explaining your lack of findings? So we did, in this next paper that was published last year, a more sophisticated examination of this question. And that QR code will take you directly to that paper, which is now, it'll be publicly available pretty soon. So in this paper, we looked all the way out to the age 25 assessment point. We looked at stimulant treatment across the whole span of time. We measured as whether or not they were treated at least 50% or more of the days in each year of treatment. And that might not sound like a lot, like if they were only treated half the time. Yeah, that doesn't sound like a lot. The reality was, if they met that threshold, most of them were treated for most of the year anyway. We looked at the frequency of heavy drinking. So we're back to that substance use consumption, not just disorder. We looked at how frequently they reported drinking heavily, how frequently they reported using marijuana, whether or not they were daily smokers, and then any of the other substances that they might have been using as an additional variable. We had two main questions in this paper. Does current treatment and treatment in the last year, does that relate to their substance use at each point in time as we followed them from childhood all the way up to age 25? And we looked at the interaction between those two as well. Because one of the things that kept happening when we presented the adolescent results was, well, what if they just went off? And that's why their substance use is higher. What about this off on? What if they just started medication? And that's why their substance use is higher. Because they aren't doing well. And they have lots of problems. And they're using substances. And now they're using stimulants as well. And those are tracking together. Got lots of questions like that. So that caused us to look at both current and last year and their interaction. Then we also wanted to address the question about, does more years of stimulant treatment make a difference? So it should, you think, if you believe clinically that it's going to reduce your symptoms, it's going to improve your impulsivity and make you do better over time. Does more years of stimulant treatment predict adulthood substance use? We did that analysis as well and used a particular type of modeling that's called marginal structural models. And without getting into the weeds of it, what's really great about that analysis is it allows you to adjust for all of those variables that might drive both treatment and substance use. Yep, five minutes? Got it. So that's what we did in that analysis. This is a really important graph because what it shows is juxtaposed patterns in the data where stimulant treatment goes down over time in the sample at the same time as substance use is going up. This is what happens naturally. It's not a great surprise if you stop to think about it clinically. I'm sure there are some of you who are thinking, well, yeah, I know that. Kids who experiment with substance use and then keep going, this is just the natural phenomenon. Use increases with age. What some clinicians don't see is that decline in stimulant treatment because the people who come to your office repetitively throughout their lives are the ones who are staying on stimulants. What you don't see are the ones who stop coming. And they're the ones who stop taking. Maybe you know they exist because they come back to you later, but what we find in many longitudinal studies of children with ADHD is the tendency for them to go off of stimulants with age. If you put these things together in an analysis, you have the potential to create a false protective effect of stimulants because you have this effect where you see, well, they're going off stimulants, but their substance use is going up. So doesn't that mean that if they stayed on their stimulants, they would be less likely to use substances? So here are the results. First, without controlling for anything, yes, that's exactly what the analysis showed, that more stimulants, less substance use. That's exactly what it showed. But then when you put in control variables, which is how old they were and the rate of change of their substance use, you lose that significance. So the protective effect actually went away when we took it to a higher level of adjustment. In part two, we adjusted for a whole suite of variables, a large number of them, that could drive both stimulant treatment as well as tendency to go get treated for substance use disorder. And we found no associations, no evidence of a cumulative effect of years on stimulants, no evidence for continuous uninterrupted stimulant medication. So we just calculated their length of periods of being stimulant treated, no association. And then it didn't matter if they were treated prior to study entry. And then we did the same thing for substance use disorder, no difference. So the conclusions from our perspective from this study clinically is that I like to say the good news is stimulants don't appear to increase risk for substance use disorder on average across the studies that have examined this question. It doesn't mean there aren't subgroups such as Ilian that is talking about. But across all the studies that have examined this, we either find protective effects or we find an absence of effects. Our study with all of these control variables found no association, so no elevated risk clinically. Could there be circumstances like what Ilian's talking about? Certainly there could be. And frankly, when we're dealing with substance use treatment, we need to think holistically anyway. We need to think about all the factors. So if you're worried about potential risk, Jeff's got a really interesting talk about sequencing non-stimulant medications and stimulant medications. Ask him about that. And think about other ways that you can address substance use protection in addition to stimulant treatment. Thank you. Thank you. Let's just see if there are any questions about this talk before we move on. I want to ask you one thing. Maybe you'll think it's better to talk about it afterwards. But I think the big question here is how to reconcile these data with the epidemiologic data, the registry data. And I don't know if you have any comments about that. Yeah. So one of the things about those data sets is they're tending to look at variables that are pretty extreme. So for example, presentation to the emergency department for substance use disorder as opposed to weekly cannabis use. Those are two very different things. And it's a pretty rare outcome to present to the emergency department or to have some vehicular accident because of substance use or alcohol. And those analyses are missing all of the individuals who are using at levels and not presenting. So that's one reason that it could be potentially creating a difference. Preventing severe outcomes, perhaps. We didn't see it with substance use disorder in the MTA. But it's also a relatively small sample compared to, hard to say that, right? It's a small sample compared to those huge epi data sets. James, we'll come back to that point. It's really great. Excellent. Well, thanks very much for being here, and hopefully you'll find this entertaining. I have been warned that this talk is quite dense. I will try and make it as friendly as possible. It is a bit dense. I was looking at it last night after I got Jeff's email and thought, oh, dear. But I couldn't make too many changes, so here we go. So yeah, I'm going to be talking to you about learning, particularly instrumental learning and striatal cortical functional connectivity in adolescence with alcohol and cannabis use. So alcohol and cannabis use disorders are the most prevalent substance use disorders affecting US adolescents today. What you see is that both have been linked to problems in instrumental learning. And one of the reasons why we might be particularly concerned about this is because at least some psychosocial interventions for substance use have, as one of the sort of implicit structures for the treatment process, a reliance on instrumental learning. So if there's a problem in instrumental learning, that's likely to interfere with treatment. We see that very consistently, high levels of alcohol use disorder symptoms are associated with very strongly reduced reward responsiveness. It came up in some of the talks, particularly with Ileane before. Reduced reward responsiveness in individuals who have engaged in high levels of alcohol use. That data is much less clear with individuals who've engaged in high levels of cannabis use. It's not that it's not there, but the literature is much weaker. And there's a few suggestions, papers reporting it going the other way. Now, striatal regions are thought to operate in concert with frontal parietal regions during the instrumental learning process. The basic, the sort of crude level. You get an unexpected reward. That's a prediction error. You were expecting nothing. Particularly, you suddenly got this reward. Striatum is interacting then with frontal parietal regions involved in attention to try and make your brain. It's trying to make sure it processes what actions to what stimuli gave rise to that reward. So that in the future, when you do that act in response to that stimulus, you expect you'll get the reward. And you don't have a prediction error. You get the reward that you were anticipating. So in the early stages of learning, we should see strong striatal activation to the rewards. And that's coupled with this sort of connectivity or certainly strong interactions with frontal parietal regions attempting to drive an attentional response to the relevant details in the environment. But we just don't know to what extent that that is impacted in individuals, adolescent individuals who engaged in high levels of substance use. There's one study in adults with substance use. That's a population with high levels of alcohol use problems. That did suggest instrumental learning problems. And it also suggested a relative failure in connectivity between striatal regions and dorsolateral prefrontal regions that are involved in attention. But that was just the only study out there. So what we were interested in doing is seeing whether the connectivity changed between striatum and frontal parietal regions changed as a function of learning in adolescent individuals and changed to a lesser extent or a differential extent in individuals who had high levels of alcohol use and or cannabis use disorder symptoms. And whether we would see a reduced rate of learning in these individuals that would relate to these problems in brain-related connectivity. And we chose to do this with this task called the passive avoidance task. The passive avoidance task is a deliciously simple paradigm. All you have to do is there is stimuli that are on the screen. And you have to decide whether to respond to them or not respond to them. Now, some of those stimuli are good. When you respond to them, you're more often going to get reward than you will get punishment. Some of those stimuli are bad. You're more often to get punishment than you are to get reward. So correct responses are responding to the good stimuli even if sometimes you're punished. And bad responses or incorrect responses are responding to the bad stimuli even though sometimes you may get rewarded. A really basic instrumental learning task. What we were predicting was that individuals with high levels of alcohol use disorder symptoms and high levels of cannabis use disorder symptoms potentially would show a slower, a weaker decrease in their error rate over time and potentially might show significantly greater errors even late on in the learning process. We were also expecting that healthy adolescents would show this nice, strong connectivity between striatal reward regions and frontal parietal attentional regions right early on when the maximum amount of prediction errors are going on, when the maximum amount of learning is going on. But then this would decline as the individuals got familiar with the task and started doing very well on the task. We didn't have a clear prediction with respect to the individuals with substance use. Two possible scenarios. One, there would be some sort of delay problem that they would be slower to start generating this connectivity between frontal parietal and striatal regions. Alternatively, what we would see would be just a failure to ever develop that connectivity between frontal parietal regions and striatal. And so we were testing those hypotheses. Population, you can't really see, but they were an adolescent sample, or I think 14 to 17 years of age. The slightly preponderance of males, but, yeah, the details are all in there. In order to be classified as having high AUD symptoms, you had to score over four, four or more, on the alcohol use disorder inventory. To be classified as having high cannabis use disorder symptoms, you had to score over eight on the cannabis use disorder inventory. These cutoffs coming straight out of the literature. So general task effects. This is for the whole sample as a whole. That's the task on the very left-hand side. And what you see is this very rapid learning. As I said, it's a nice, easy task, relatively speaking, for most people. Relatively rapid learning. So you have this, for the first half of trials, it was quite an easy cutoff. For the first half of the trials, you get this very rapid decrease in the rate of errors, and then it sort of flatlines as the individuals basically learn the task. And if you break it down, you see that really rapid decline in the first half, and then the flatline after that in the second half. Now, if you look at the patient samples, what you're seeing, so on the far left-hand side, is the individuals with low alcohol-use disorder symptoms and low cannabis-use disorder symptoms. They're showing this really rapid decline in error rate, and they're showing a very low error rate at the end of the task. Both individuals who had high levels of cannabis-use disorder symptoms and those with high alcohol-use disorder symptoms were slower to show that decline. They were reduced in the rate at which they declined their error rate. And in fact, those individuals with high alcohol-use disorder symptoms, those individuals were still showing significant levels of impairment even late on in the task, relative to the low-use individuals. So there was quite significant impairment behaviourally on the task, particularly for the individuals with high alcohol-use disorder symptoms. We did a very basic contrast, reward versus punishment, to identify regions within the sample as a whole who were responsive to reward. You were responsive to reward. And these are sort of classic regions. I can't have a pointer, but the classic striatal effect, probably easier to see here, actually. So we had a striatal region, that region in yellow, which was our core seed region for the connectivity analysis. And then we had two sets of regions that came up through the contrast. These were frontal parietal regions. Those are the ones in red that you can see in the image. And then there were other non-frontal, non-tensional regions, the ones in blue. They were sort of control regions. We were assuming that our learning rate phenomena was going to be very clearly, connectivity changes were going to be occurring within the frontal parietal regions as it was interacting with striatum, not between the non-frontal parietal regions. And that's basically what we get. So if we look at the blue line, that's the individuals with low alcohol, low cannabis use disorder symptoms. They show this very nice, just as we were hoping to see, really strong connectivity between striatum and frontal parietal regions early on in the task that declined as their learning became successful. So they were consistent with the idea that you're getting these surprise reactions, organizing attentional response, focusing in on the details of the stimulus and the details of the response. You learn exactly what to do in order to get that reward and not have things go awry. We don't see that in the individuals with high cannabis use disorder symptoms. And in fact, they were seeing, if the easiest way of imagining what we're seeing here is some form of delay in that increased response. They're not showing it early on. In fact, even late on, they're finally starting to interact between striatum and frontal parietal regions. There were no phase effects for the striatal interactions with the non-frontal parietal regions. This is very specific to the way that striatum interacts with frontal parietal regions to drive attentional responses. With respect to the alcohol use disorder group with high alcohol use disorder symptoms, here we didn't see such a dramatic effect except for the individuals with high alcohol use disorder symptoms just didn't show any changes in connectivity strength across the study as a whole. They basically were flatline. I mean, there is not a flatline on the image, but it was a non-significant increase. They weren't showing any change across the experiment as a whole. So, getting on to the conclusions. First thing to remember that we're seeing within individuals with high levels of cannabis use disorder symptoms, high levels of alcohol use disorder symptoms, indications of instrumental learning problems with the implications of that with respect to psychosocial interventions. It's definitely more marked in the individuals with alcohol use disorder. They're not just slower to learn. They stay disrupted. They stay poor performance for longer than the individuals with cannabis use disorder symptoms. But both groups are having, or at least it's a more general problem. It's not just confined to cannabis use disorder. Alcohol use disorder. With respect to the low disorder symptom individuals, the typically developing in the loose sense of the word, this was not a very typical developing population. There was a lot of psychiatric comorbidity in this population, but still, at least they were not showing these high levels of alcohol use and cannabis use disorder symptoms. In this sample, they do show this marked, strong connectivity, marked strong interactions in the early phase of the study, just as we were expecting, that massively declined as they became successful. When they needed at the beginning to respond to that reward, generate an attentional response, get organized with respect to coding, what is the good things in the environment? What are the good behaviors in the environment? They were successfully doing that behaviorally, and they were successfully doing that with respect to the connectivity, and that was declining as their task performance became at a sort of asymptote level of success. With the individuals with cannabis use disorder, we see slower learning of the contingencies relative to the low use group, and we also see this very odd pattern of a delayed increase in striatal connectivity with frontal parietal regions. And to be honest, it's not completely sure what's driving that. One possibility is that we're just seeing the slightly disruptive, there's some indications that cannabis use may disrupt dopaminergic signaling less than alcohol use. It's not clear, but there's some suggestion. And we may be seeing less of a problem in this sample because there just is less of a problem. The other possibility is that one of the things we're seeing is actually interaction with problems elsewhere, problems in attentional systems more generally, which is what we see often reported in individuals who are showing high levels of cannabis use disorder symptoms. What we're seeing, though, in alcohol use disorder is a very much more marked behavioral problem. They're significantly impaired, they're slower to learn, and they don't really learn terribly well. And they don't really learn terribly successfully, at least not relative to typically developing individuals. And they're not showing ever this change, at least within the scope of the paradigm that we were looking at. They don't show ever this increase in connectivity between striatum. They can't recruit that interaction between, or can't do it to a lesser extent as a group, the interactions between striatum and frontal parietal regions. So, healthy participants show rapid learning, or participants show reasonably rapid learning on the passive avoidance task. And this is accompanied in the healthy individuals, or the low-symptom individuals, rather, by initial high connectivity between striatum and frontal parietal regions that then declines as successful learning occurs. Cannabis use disorder, high cannabis use disorder symptom groups, the learning is slower. And in contrast to the low disorder group, they show an increase, or a delayed increase, in that striatum frontal parietal interaction. But individuals with a high alcohol use disorder are really more significantly still impaired in their behavioral performance. And on top of that, they don't ever show. It's all flatlined. There's an absence of change in connectivity strength across the learning experience. Jeff, also, I made sure that I put in some clinical implications. So, I have some quick clinical implications. With the big caveat that I'm not a clinician, so I always feel a little bit sheepish whenever I'm doing this, so don't lynch me. So, here we go. I think there are some very clear, the really very clearest one is that, you know, I came up a little bit with respect to your talk as well, thinking in terms of what difficulties an individual case shows when they're in an intervention based around what they can and cannot do. If they do have, and again, we came up again also with your talk particularly, this issue, and also with yours as well, this issue about group level analysis versus individual strengths and difficulties. All of the data I was showing you were groups of patients. But not all of those individuals were showing the same level of impairment. Some individuals in that group were showing quite profound levels of impairment in the learning experience. Those individuals are likely to significantly struggle in psychosocial interventions that are reliant on a degree of instrumental learning, particularly anything reward-based. This, when I was doing this, this was done at Boys Town in Omaha. And there, that's a heavily psychosocial intervention with a lot of reward-based interventions to it. You basically have to make sure, in order to generate interactive strength between the conditions and the individuals there, at least sort of a four to one ratio of good experiences versus one. So if somebody has to be told off for doing something, you've got to make sure there's at least four positive interactions that are bracketing this one negative interaction. Potentially, that type of psychosocial intervention has to be ramped up in individuals who have these types of problems and these types of reduced reward response. There's some implications with respect to ADHD. I mean, if there really are these compromised dopaminergic signal, if substance use, particularly alcohol use, either is a risk factor or alternatively, as I think some of these things we're seeing are, as an impact on the brain of chronic use, heavy use, in particular alcohol use, you could imagine situations where that's going to interfere, not just with, it would interfere with treatments that are reliant on dopamine. And certainly with respect to, I can't talk for the frontal parietal attentional components of the impact of methylphenidate, but certainly with implications with respect to reward response that we've looked at in some other studies, there you can imagine that if a patient has problems in generating reward responses, partly induced by the fact of chronic alcohol use, that individual is not going to benefit from stimulants as much as another individual might. And the final one, this is my home zone, the implications for assessment. It sort of came up in several of the talks. The fact is, if it's true, if we really are finding that if you have an instrumental learning problem, you are significantly less able to take advantage of some psychosocial interventions. Theoretically, we expect, but we have not documented that to be the case. If that is true, we need markers. We need markers. We need a signal that tells you this individual is a really good responder and that individual is not a good responder. I am not a fan of biomarkers for diagnosis. I really don't. I think we've got diagnosis tools already. We have questionnaires on top of the diagnosis tools. We've got that. I mean, we have worries about reliability, but I don't think those issues are ever gonna be solved by biomarkers. But we really do need them to be able to get individualized care, to say this individual, right in front of me right now, has a real profound problem in instrumental learning and is not gonna benefit as much in this type of intervention as somebody else will. And when we have those things, then we can, at least if we find that they actually do make these sort of predictive impacts, then we can start. Do I have two minutes to say some extra bits of data that come from something slightly different? I don't, all right, fair enough. Well, it'll depend on all of you. If you have questions, you won't have time. And it'll be okay, because we really wanna get to those. And if you don't have questions, we will fill in, don't worry. I want you guys to come up, Brooke and Ilyan. Yeah, I have questions too and comments to make. I want, we also have people watching remotely and they're gonna, I'm just gonna make a couple of comments while they start in their questions and I'll read them off, assuming there are some. If not, we'll have a panel discussion here. James, fantastic for a non-clinician. Fantastic, you got really where that thing needed to go. Loved it. Three really great talks, right? By the way, can you just, I think you made it clear at the end. What is instrumental learning for anybody who, these terms, by the way, are very confusing. They're not user-friendly. Sorry, goodness, that's loud. Instrumental learning is just learning behaviors in order to get rewards. I mean, there's more, but in this context, it's like my task, responding to the good stimuli in order to get reward, avoiding responding to the bad stimuli in order to avoid punishment. Exactly, so for gain, for benefit, right? It's funny, before you got to your last point, I went over to Brooke and I said, this is really cool because if we had a predictor of who could use reward-based behavioral treatment, that'd be really terrific thing. That'd be a great way to use a paradigm like this and then you got there without us. That was good. Really, really enjoyed that. I did wanna say, James, that I don't think you told us about the rates of ADHD in the sample. There's a possibility that they could differ across your groups or that it could somehow or other moderate the learning response. I wanna ask you about that. So, it would be pretty constant in the two high-use groups. It probably, it almost certainly was lower in the low-use group. I can't, I mean, I know it's in that table, but I can't remember the exact numbers. By the way, in Ilian's talk, there's also another term that is, I think, not user-friendly, which is, well, there are three. Unexpected non-reward, frustrated non-reward, and surprising non-reward. Maybe they're a little easier to figure out, but I just wanna point out, as a child psychiatrist, you deal with this all the time because what happens? Who are the kids that have these tantrums? The kids who think they're gonna get a reward for something and they don't get them. And so, that's what's being modeled, really, in these kinds of paradigms. So, I just wanna make that point. And, okay, let's see what the group has to say. Do we have any questions? We may have questions over here. Good, because the audience, I'll monitor. Go ahead. Pietro Pietrini from Italy. Nice talk, solid talk. I just want to follow this. You said you are not a fan of biomarkers, but as a matter of fact, biomarkers in psychiatry are much needed to make psychiatry closer to medicine. And I have a question for you. From a forensic point of view, when you are called to examine young adults for mental insanity, even for minor crimes, but these people very often commit minor crimes, sometimes even bigger crimes. I wonder if this, what you showed, could be used in order to claim mental insanity. Because, as you know, in Italy, for example, in Italy, in order for alcoholic use or substance use to be a valid reason for mental insanity, you have to show chronic use and damage. But, when you don't have this, but you have something like this, but you could show that we are difficult learner. I mean, we don't learn from experience. And we do have a different frontal parietal connectivity. What do you think about this? Is this something in the future we can use, in the near future? Thanks. So, I'm, sorry, I did not mean to say I was not a fan of biomarkers. I'm a deep and in love passionately with biomarkers. I just. I know your area of work. I mean, I know aggressiveness. I just, for diagnosis, I find them, there's not the way that I'm pursuing things. I do find them really interesting with respect to treatment response and with respect to understanding an individual's problem. So, back to your, the second half. Yes, indeed, I mean, so I can't say about the Italian process as regards making the, but could we have tasks that would identify a problem? Yes. We don't have them. And one of the problems we've faced with the neuropsychological tasks has been a problem on reliability, that they've got low reliability. Having said that, that's almost now being indicated that it's primarily a weakness of some of the designs that have been tested. There are ways of dealing with that problem that actually remove those reliability, at least anything, I mean, up to the 0.8, 0.9 reliability, which may not be as good as some things we might have, but you can still get very strong reliability if the right designs are used. So whether this will be the right type of design, I can't say, because we haven't done that, but there's certainly, you can find the tasks that will do it in some contexts. And again, I think with respect to the neuroimaging, I mean, there's all sorts of problems with respect to using it as a marker, but if we could show that we have a reliable marker, and we could show the extent to which it's disrupted relative to the healthy samples, so if somebody is significantly impaired, then at least we could draw some of these sorts of conclusions. Thank you. Can I just add one? So one of the struggles with this, at this point, is the vast majority of the tasks that we use don't relate well to daily life functioning, right? So we develop these tasks so we can have a very discrete behavioral measure and put somebody in the magnet and measure their brain activity and connectivity, but then when you try to look at the correlation between their performance on the task and their functioning in daily life, it's almost always abysmal. But I do think, right, it's getting, I mean, there's an attempt to try and improve that, right? I think there's two, yeah, there's the problem, if you don't have a reliable task, then you will never get good correlations. Reliability first. Yeah, exactly. So, so. But even that, I think once you get a reliable task, it's such a discrete, fine, precise measure. I'm still not convinced. So I think it's difficult. It depends on how broad the behavioral paradigm is. You know, how broad the behavioral thing that you're looking for is. I mean, I think, you know, I think one of the, if I was gonna be developing a marker that would do prediction, would I do one that would predict use of alcohol? Probably not, because I think there's so many variables that the brain is not that important, I mean, relatively speaking, amongst all the social and other one. If I was going to get one that was predicting the types of thing, you know, like the drink driving and, you know, doing all that, then I think you could actually, you're more likely to succeed. So if the functional skill level is gonna have a real impact on an individual's existence, I mean, but you have to accept that. Learning to drive is a good one, right? Well, there's a lot of, there are a lot of neuropsych skills that are being used in learning to drive. And I mean, IQ is a clear neuropsych task, and that predicts everything. I mean, lots of things, anyway, so. Not everything, not everything, I do, I was hyperbola there, but it definitely has a big predictions on many aspects of our existence. So I think it's partly because there's been a few very tight neuropsych tasks from neurological patients applied to ADHD, where it's particularly not been successful. But I think in other areas that may not be quite so, and again, if they're reliable. Do we have other questions? Just step to the mic. If you don't step to the mic, I won't know that you're ready to ask something. So come on up. Go ahead. Oh, hi. Thank you for the great talk. I have a question for James. So I was wondering regarding the difference in instrumental learning. How do you make, so how do you control or make sure that the motivation level are on par? So in line with that, I'm curious, how would you gather these group of adolescents with a substance use? And how do you make sure that they join this study with the same motivational level and willingness to do it well? And what kind of approaches that you use to kind of control for that as well? Yeah, I mean, I think partly we can't, I wouldn't be at all surprised. We've got a population potentially that are not so reward responsive. And just like Ileane and you were saying as well, if you've got problems in that reward response, then you're not going to be so motivated to engage. Now, having said all that, the kids, this was primarily from, they were, this was work when I was at Omaha in the boys town, a boys town. And there they were in residential care and they had the opportunity to just come out and play in the scanner and do the tasks and do something different. So sometimes they didn't want to, sometimes they did want to. And it's certainly, they engaged in 10 different imaging paradigms and a bunch of neuropsych tasks as well. And it's certainly not that you would find this little kiddie over here who flunked the lot. There was a lot of, most individuals were successful in some areas and failed, well, failed or didn't do so well in other areas. So do I think that it's completely not a problem? Absolutely not. But I think that's because that is the pathology. But do I think that it's the, they just were staring into space and doing nothing? No, because they were engaging in many of the other tasks as well. And as you can see, it's slower. I mean, apart from some of the individuals with high levels of alcohol use who really were still showing quite profound, but it's still slower rather than flatlining, at least in their behavioral performance. Next. Along with that, so did you do a post where people had treatment for three months, six months to do again? Because this is a population I work with and most of the time we don't do any testing for them because it's inaccurate. But it's good to see that because they have a lot of behavioral programs that they fail and everyone wonders why they fail because they're not really learning. But with three, four months sobriety or with significant decrease, did you do a sample with that group? So, well, what we'd had hoped, the original design of that study was that we were going to see the individuals as they, or at least what we came on to design, we've seen them as they came in. The treatment program is usually about a year long time in the residential care of Boydstown. And then the individuals go back to their homes or they go on to alternate, a lot of them go into the army and things like that. And so we were hoping to see them as they came in, as they were about to leave and then follow them up afterwards. But COVID blew us completely away. And so we just started collecting. We did in the first year, we weren't going into doing the time two. We then were sliding into time two. And then after I think the 67th time two person, COVID came and we shut down and started analyzing data. So I completely agree. It would be really interesting to know. Yeah, I will say that just because I'm going to take this, I apologize. I think you have time for your two minutes. Thank you. But just because I just got the data literally two days ago, it was deeply entertaining. This is not substance use data, but it's conduct disorder data. And there's little kiddies, but it is with that passive avoidance paradigm. And there's a psychosocial intervention that they use in the Institute of Psychiatry in London, parenting intervention, that has a relatively high success rate. I mean, it's 50%, but that is pretty high for conduct disorder success rates. But it meant that we can look at successful responders versus non-responders. And what was dramatic for the passive avoidance paradigm, where there was a lot of problems, particularly and even more striking problems on this paradigm in conduct problem individuals and in individuals who are getting their individual substance use. We saw that in the responders, you got significant improvement in passive avoidance performance. And in the responders, you also saw recruitment of some of at least the ventromedial prefrontal cortex and posterior cingulate cortex. We didn't see a striatal change, but we did in the other two regions. So this possibility that there's some degree of rescue in a slightly different phenomena and in treatment responders. So there's all sorts of caveats there, but I found it so deeply entertaining that, you know, and also optimistic, because it, you know, I mean, unfortunately, there were some of the individuals who were not being helped and who stayed quite disrupted, but still, we did see some change. Hi, I'm a child psychiatrist working in community mental health. So, and I did arrive a little bit late, but is there a protective effect of treatment in terms of treating kids who have ADHD in terms of possibly, I guess, mitigating the risks of substance use? Is that something that can be taken away? I was a little disheartened by your data showing that, I guess it's positive that stimulants don't seem to increase the risk of substance use, but also that it doesn't seem like there's really a protective effect. So I'm wondering kind of what can I do? What can I tell families who may be on the fence about medication? If other sorts of treatment modalities may be protective to help kids. Thank you. Brooke, why don't you take that? How many of you would launch a stimulant treatment in a child specifically for the purpose of protecting against substance use disorder? Yeah, okay. So that's a little bit different. That would be kind of starting later, right? And I think... Well, a parent might be more willing to consent if they thought that it could mitigate a later risk. So some parents might say, I don't want treatment, but if you say, well, there are these risks that could develop later, a parent might be more willing. I mean, in areas where there's already impairment, I wouldn't just, I'm not suggesting just to start it as a test, but that might be a factor that a family would consider. That's helpful to hear. So, and I didn't mean to challenge you. It's more like just kind of interest. The answer is truly gonna depend a bit on who you ask. All right, what I will say confidently is the clinical data are not showing increased substance use disorder risk from treatment. I feel confident in saying that. There are large epidemiologic studies that show that when individuals have been treated, they are less likely to present to the emergency department for substance use disorder. They are less likely to have, forget what the other outcomes are in those studies, but they're pretty extreme. It's like treatment for substance use disorder, presentation for, yeah, different kinds of incidents. Yeah, yeah, yeah, like car accidents, that kind of thing. What those big, huge studies haven't done, because they didn't have the measures, was they weren't able to look at things like using the substances themselves. So that's my concern about those studies is that they just don't have frequency of use, quantity of use, those kinds of indicators. So I think they're missing that information, but. I would say that it's, they have the advantage, though, of having very big samples. Absolutely. They're huge. They're huge. They're, like, countrywide, you know, national data sets. They're enormous. From medical registry, like, in, yeah. I want to pick up Brooke's question, because I think it's a really great question. Would you, and my answer would be no, also. I would not, I wouldn't start someone on stimulants to prevent substance abuse later in life. Why? Because I initiate treatment almost always for more proximal outcomes. You don't think about distal outcomes. And it's the difference between what you do in practice and how you interpret it over time and in a population. So stimulant treatment prevents a lot of bad outcomes over time. You could think about, you know, treating ADHD in terms of, you know, more widespread prevention of comorbidity, you know, prevention of depression, prevention of a lot of other things. There's nothing wrong with telling people that there are a lot of long-term risks for, that are associated with ADHD that are improved by stimulant treatment. But I would tend to agree with Brooke, though, that most people will want to see something more immediately tangible when they're making that decision. Yeah, and I think it's also, let's look at the immediate moment, you know, and how severe is the symptomatology? What are the impairments that are happening right now? And let's think about the next year. And what is your child's life gonna look like? What is your life gonna look like over the next year in the absence of treatment? And it can be an experiment. It can say, be six months, let's try. But the beauty of, at least for stimulants, right, is their ability to be acutely, immediately efficacious. And an experiment can be undergone. And the effects can be so powerful, the people will have used that experiment to decide, right? And then I just do think it's a very heartening finding. I'm a psychologist, so I'm not prescribing. But it is a heartening finding to know that we don't see these adverse negative effects. So I think those two things paired together can create a pretty compelling clinical picture. I mean, what percentage of your patients, when you talk to them about using stimulants, ask, will it carry risk for substance abuse? I think it's a high number. I mean, it is a high number for me. And so it's actually really important to be able to answer that question. Well, they don't, they, I've had some parents ask about risk of, yeah, substance use disorder on the medication, but unfortunately, I ran into a number of families who are not comfortable with starting medication. So I think, you know, I agree about the immediate effects. This isn't an acute need program, but also that it's not just about the present, but that there can be, as you're kind of saying, longer term, potentially serious outcome of untreated ADHD over time, which could involve risk-taking behavior. So I was more trying to, I guess, frame any of the data supports, I mean, I guess, mitigating the risk of substance use disorder, whether or not it's- Steve's paper. Steve's paper, the review paper. You know, that Steve Ferron, that review paper that I meant, I don't know, did you see it? I think I took a picture. Yeah, that will be helpful to you. Thank you. Yeah. It is great, that international, the thing that's also amazing about that paper is that it was really a consensus across the world. You know, people who are expert in this area worldwide coming together and doing this. Steve led it when he became president of the World Federation on ADHD. It was one of the really important things he did. And thanks very much, those of you who stuck with us, and for your interest and your attention, and hope you enjoy the rest of the meeting. Thank you. Thank you. No question.

substance use disorders

youth

disruptive behavior disorders

ADHD

conduct disorders

stimulant treatments

neurobiological predispositions

psychosocial risk factors

MTA study

alcohol use disorders

cannabis use

instrumental learning

personalized treatment