Latest / The Biological Psychiatry Podcast / Differential Effects of Psychotic Illness and Antipsychotics on the Brain
Transcript
- Elvisha Dhamala: From the Biological Psychiatry family of journals, this is the Biological Psychiatry podcast, ⁓ where we discuss the latest research on psychiatric disorders and their treatments. ⁓ Thank you for joining us. I'm your host and Social Media Editor, Elvisha Dhamala. ⁓ When we see changes in the brain in someone with psychosis, is that the disease itself that's causing it or a side effect of the medications we're prescribing to treat it? This question has been surprisingly difficult to answer because almost everyone who develops psychosis is immediately started on antipsychotic medication, making it nearly impossible to separate what the illness does to the brain from what the drugs do. Today, we're discussing a study conducted as a collaboration between Origin, the University of Melbourne, and Monash University that tackled this very question studying people experiencing their very first episode of psychosis who had never taken antipsychotic medication before. The results challenge some widely held assumptions and offer a new perspective on what's really driving brain changes in early psychosis. Joining me is Dr. Sid Chopra from Origin and the University of Melbourne, first and corresponding author of a recent paper in biological psychiatry titled Differential Effects of Illness and Antipsychotics on Cortical Thinning in First Episode Psychosis. Dr. Chopra, welcome to the Biological Psychiatry Podcast. So let's get started. Can you tell us more about why this question of whether it's the illness or the medication driving brain changes in psychosis has been so difficult? Sid: I think the the primary reason is that there's a confounding between medication and the illness severity, right? It's called confounding by indication in a way. Like people who have more severe illness symptoms will be given more medications, well I guess high doses of medications for longer. And this kind of confounds illness severity and medication dose and Antipsychotics remain the first-line treatment and they kind of have since maybe the 50s. and it's only after that time did kind of modern imaging technologies like CT and MRIs really become widespread. So all these kind of hundreds, thousands of imaging studies we've done, the people who are in them have all almost always been exposed to medications. And if they have not, it's only kind of in single time point, Just before they start their medications. So It's made it incredibly difficult to disentangle what brain changes are due to illness and what brain changes are due to medication. This is a really important question, Because if these brain changes are really an itrogenic and harmful effect of medication, that's something we need to know as soon as possible. Elvisha Dhamala: In your paper, you mentioned that there's been some animal studies and large observational studies pointing to kind of both directions of this question. So some suggesting that antipsychotics cause brain tissue loss, others suggesting that they might be protective. Can you tell us a little bit more about what we've known about this topic before this study? Sid: Yeah. I think that the the evidence has generally leaned towards medications having iotrogenic effects, like negative effects on brain volume and brain tissue. And the bulk of that really I think started with animal studies in the nineties and early two thousands. There's a classic kind of Dolph Peterson paper in neuropsychopharmacology. I think it was in rodents they showed that when you give rodents antipsychotics, a different even first generation, second generation versus vehicle I guess placebo sham, you do see loss of maybe it was brain weight on cell sizes and things like that in the animals. And actually it was a macaque study, if I get it right, that that showed this. And then similarly David Lewis's group showed lowering of like brain weight as well in animals given antipsychotics and this is really triggered worry. And this is one of the reasons actually this particular study had imaging in it, was because Pat McGorry designed the study, was kind of worried about these medications causing harm. The human evidence really is a couple of large longitudinal studies, Hoet al, Nancy and Dreyson's group showing that people who receive higher doses of medication show greater decline in brain volume, cortical thickness, at least using MRI Those studies are almost always, as I said earlier, confounded by indication, You can covary, of course, people add these covariates of medication exposure, but you can't covary out a causal effect. especially not a linear one, it's quite difficult, So regardless of covarying for medication dose, these studies have been confounded as as our kind of paper here shows that it does make a difference whether you give someone medication or not. Makes a huge difference. So that was kind of the evidence, I guess. for iatrogenic harm in a way, And the evidence against it has been more in terms of because it's a very hard thing to study in give placebo, it's been more in terms of first generation versus second generation, where large scale studies have shown that second generation antipsychotics may be causing less brain changes in a way, They may be leading to less decline and things like that. I think it was Jeffrey Lieberman's study, which was very large sample of of people with first lipsid schizophrenia showing that when they received a second generation antipsychotic versus I think caliperidol, there was there was less changes in the brain, And that's the idea that maybe these second generation medications might be more, I guess, protective in the sense of MRI based measures. In terms of animal evidence of protection, I don't don't know if there was really that much. There's been a lot of mechanisms that have been pitched, reducing inflammation, increasing synaptogenesis, but these have always kind of been studied in RAP models and have been kind of really difficult to understand molecular mechanisms in humans around protection anyways. Elvisha Dhamala: So this study uses a triple blind randomized placebo controlled trial where patients with first episode psychosis received either an antipsychotic or a placebo. Can you tell us a little bit about the practical challenges of running a study like this? Sid: Yeah. I mean, before I do that, I just want to say that I did not run this study, It was a large team. It was led by Shona Francy, Brian O'Donohue, Pat McGorry, Kelly Ellett leading the cognitive part and Alex Fonito kind of really leading the imaging acquisition. I did collect some of this data as I was going through my PhD and the privilege of kind of analyzing it and working on multiple kind of arms of it throughout. I guess after it finished. But through that, you know, I did get exposed to learning about the difficulties and challenges around that. I think the most practical challenge was getting people in that early stage right before they start antipsychotic medications can be difficult. But also the fact that it's a randomized controlled trial, you know, people don't want to leave their treatments up to chance. that was kind of difficult. The other challenge was also I think finding placebo drugs for the study was a little bit difficult. The pharmaceutical company providing Placebo pulled out, which is the reason that we changed from Respiridone to Paliperidone halfway through the trial, to because that was where the matched placebo availability was kind of there. in terms of actually the ethical considerations around this study, there's a really good paper by Shona Franci, who was kind of the lead author on the clinical trial that this MRI study was embedded in. It's called, I think it's called like who should be taking antipsychotics. But it's a really comprehensive Elvisha Dhamala: I can add stuff. Sid: coverage, of the ethics around withholding antipsychotic medication, how to do it ethically, and why this particular type of trial was needed at the time and and currently. Elvisha Dhamala: So you chose to look at cortical thickness rather than surface area or volume when you're looking at the brain measures themselves. What made cortical thickness the more informative measure here, and what how does it differ from what the other measures are telling us? Sid: So we have published another paper with this study using volume. So those are the kind of the two pre registered outcomes here cortical thickness and volume. And we already published a paper on volume. And briefly, I guess what we found there is antipsychotics are expanding the palatin. So in the antipsychotic group, you see an expansion of the palatin compared to the healthy controls. And in the placebo group, you see a decline, So the idea is that the antipsychotics might have been preventing this decline. But this expansion of basal basal ganglia due to antipsychotics has been shown repeatedly in the literature. I mean, we're the first to show that it's really definitely a medication effect, and the first to show that if you give placebo, you actually see a decline in the same structure. but we don't really see large or significant, statistically significant changes in the cortex, right? And the reason we chose cortical thickness is because it's more sensitive, When you look at Alzheimer's disease and other true neurodegenerative conditions. Cortical thickness always comes out as the more sensitive marker compared to volume, especially these free surfer algorithms to separate and partition these tissues get more and more advanced. You can get more and more accurate, you don't need manual intervention. and the other reason is that it seems to have a different genetic mechanism, at least like cortical thickness does, compared to volume. And the genetic mechanisms controlling cortical thickness for various reasons might be more relevant for developmental and neurodevelopmental disorders like schizophrenia, And we did look at surface area. It's in the paper. It was more of a side analysis. and we didn't find anything there. And I guess the classic way to think about it is that surface area is more of an early stage developmental marker, whereas cortical thickness is more susceptible to environmental effects and things like that. And here we're really looking at medication effects, in terms of causal manipulation, which is why we thought cortical thickness again would be the the better variable to look at, and it did turn out to be the variable that was most impacted here compared to surface area. Elvisha Dhamala: So now let's get into your results themselves. You show that patients who received placebo showed widespread cortical thinning over 12 months compared to healthy controls, but the patients who received antipsychotics did not. Can you tell us how exactly you determined this and what the magnitude of thinning looked like? Sid: Yeah, so we kind of took two approaches to this analysis. And the first approach was to run kind of pre-specified contrast. You know, when you have three groups, three time points, we kind of chose and this is what we did in our previous study looking at brain functioning in the same cohort. We pre-specified two contrasts that kind of increases your power, Because there were certain comparisons in that three group by three time interactions we were not interested in. So we we did what we did last time and we looked at The first contrast is kind of the illness effect, Placebo versus healthy control. And here we see widespread thinning, as you said, There's certain areas that are more preferentially impacted, but generally you see it a brain-wide effect of thinning, So that's less cortical thinning, less cortical thickness in the placebo group compared to the healthy controls, right? When you look at the antipsychotic effect, that's what's medication doing to the brain, you don't see this level of thinning. Actually, you don't detect anything statistically significant. That was the first way of looking at it, the second way of looking at it is to actually just do a full-blown interaction effect where you have all three groups in the model across the three time points. And when we run it that way, we kind of come to the same conclusion. The placebo group is significantly different to the healthy control group, but the antipsychotic group is not, So that's kind of how we came to this conclusion. an important point is that we're looking at it at a vertex-wise level across the brain. So at each point of the brain, we're fitting these mixed linear models that are trying to ask the question of is there differences in trajectory between these groups, Is the group or the group of people that given placebo showing difference in trajectory across the year compared to healthy controls, Who don't have this illness and didn't didn't receive placebo or medication. So kind of linear mix effects models at a vertex level would be the succinct way of saying how we actually answered this question. the magnitude of changes that we saw was approximately three percent, right? And there's a confidence interval around that. So that means that compared to the healthy control group, over the first year of illness, there was a three percent decline in the placebo group, I guess there's this trend in modern imaging and cognitive neuroscience to look at a spatial pattern and see whether it correlates with normative features from other groups or other modalities and data types. So I think we we can get into this later. But just looking at that pattern alone, you know, it doesn't necessarily just by looking at it tell us much because you see kind of sensory areas involved as well as higher dissociation areas. To really understand it, I think you need to look at spatial covariance with these normative features, right? So do they correlate? There's a map of differences that we're seeing here correlate with these other normative features from other groups, ideally not normative, meaning ideally from other patients, ideally from the same patients, but we only collected MRI data here. so when we looked at whether these this spatial pattern had similarities to other things that we know about the brain, such as functional networks or where certain cell types are, we did see associations. So I mean yeah, Elvisha Dhamala: Thanks. Sid: in terms of like what this pattern tells us, I think on its own it doesn't tell us enough, which is why we ran some of these later analyses looking at associations or correlations of what we saw with normative features with the brain, the things that we already know about the brain. Elvisha Dhamala: So the antipsychotic group showed no significant thinning relative controls, which, as we've talked about, might be counterintuitive. But you're very careful to explain in the study that a null result isn't the same as evidence of protection. Can you explain that? Sid: I think in in some ways it's not counterintuitive, Because there has been a quite a few studies that have shown a lack of cortical changes, right, in volume or thickness longitudinally in early psychosis. And most of these all of these cohorts have received medication, So potentially one of the reasons they haven't seen large effects is because these people are receiving medication, I think that's a very important point that a lack of association doesn't mean there's a lack of evidence for association. think we need to be important here because the sample sizes are small. They're not huge, right? In the antipsychotic group, by the time we get to our twelve months, we kinda only have thirteen people left in that group. And we ran a variety of different simulation analyses in the paper to show that this probably didn't impact our results too much. But nonetheless, it's it's still small numbers that we're working with here. And this is important because If we found that both the antipsychotic group and the placebo group differed from healthy controls, we could perhaps make the claim that medication is protective, The effect of medication on the brain is different to that of placebo. But here what we see is only the placebo group shows a significant difference from the healthy control group. The medication group does not, So if you imagine kind of three lines, the healthy control group being straight and the placebo group showing a decline. The medication group is kinda in the middle, And there's a figure in the supplement that shows that very clearly. that that seems to be what's happening. So the medication group's somewhere in the middle, but it's not significantly different from either. So we can't actually conclude that there is a protective effect, The only thing we can conclude is that if you don't give someone medication, they will show decline that's detectable, And widespread. And if you give someone medication that that decline isn't passing our significance thresholds. That's I think a careful way to word it, but I think I probably hopefully word it a little bit more clearly and carefully in the paper. Elvisha Dhamala: You also found that there weren't any significant changes at the three-month time point in either of the patient groups. What does that tell us about the timeline of these brain changes? Sid: So I guess what I said earlier was that when we looked at the volume effects, we do see effects at three months, but they're very clearly medication effects, right? They're really the basal ganglia, the dopaminergic regions that are expanding. Medications really are dopaminergic medications in the most part. I guess what it means is that these kinds of cortical changes accumulate over 12 months. You need longer than three months to really detect them, They're not rapid changes. I mean, some would say that. approximately three percent change every year is actually quite a large magnitude, people would argue, which is fair enough. And I don't think this three percent is consistently held throughout the illness, 'cause that would that wouldn't make sense. That would be greater atrophy than than Alzheimer's. But I guess that's the main takeaway that it it's it's really takes longer than three months to detect these thinning effects, They happen over the first year of illness. It's not necessarily the first year of illness, it's really the first year of treatment engagement because while these people were very close to illness onset, they were within a few months of illness onset, you know, it's really the first year of treatment engagement when they came in touch with with treatment services. Elvisha Dhamala: Earlier you also mentioned some additional analyses you did to try and understand what might be driving these. So you looked at specific molecular and cellular patterns, and you found that the spatial pattern of thinning aligned with the distributions of several neurotransmitter receptors. Can you tell us a little bit more about that? Sid: So this is what I was referring to before in terms of the trend in modern cognitive science to look at spatial associations between these MRI maps where we really have no molecular or cellular information and see if they correlate with other modalities. So here we got these receptor maps for neuromaps, which is a toolbox that's kinda collated pet tracer binding. I guess normative maps of healthy controls that have been averaged across different pet metrics for different receptors. And we looked at the correlation between what we saw in terms of the effect size map, Which is the difference between people who received placebo and healthy controls over time, the difference map and these pet maps. And we saw that there were correlations with different serotonin maps, histaminergic maps, and I think GABA A at the benzodiazepine binding site. I think a good way to think about these kinds of analyses. Especially with schizophrenia, because almost each of these receptor systems has been implicated in schizophrenia, because we don't know what causes schizophrenia. It's a variety of diseases most likely. I don't think we need to go into individual receptor stories here because this is a story with every receptor with schizophrenia. What's more important I think is that these sorts of analyses validate that what we're looking at in terms of our MRI pattern is related to the brain. When I say that what I mean is that there's a lot of artifacts that can be here, They can be why are we seeing this effect? Is it because of dehydration, because of antipsychotics, right? Mental and physical stress, Head motion, there's a lot of different artifacts that are at play with MRI. But when you start seeing spatial correlations with receptors, you can I guess in a way it gives you evidence that maybe what you're looking at is neurons, These receptors are largely expressed on neurons. So it's a way of validating that we're looking at the neuron or a neuronal effect, Rather than an artifact or some other compartment in the brain. So that's kinda kinda how I think about these receptor analyses. Elvisha Dhamala: You also found that the most highly differentiated primary sensory cortices were relatively spared from thinning. what might explain why these regions are more protected while the association cortex is more vulnerable? Sid: So that was our hypothesis going in in a way that the side of pathology and schizophrenia is in the association areas. This has been shown by so many different people in so many different studies, but you still see actually other studies implicating sensory cortices. Once again, this huge heterogeneity depending on study and modality that you're looking at. But what we found is that this thinning that you're seeing, right, across the first of illness when someone doesn't receive medication, right, this kind of likely illness effect that you're seeing. It doesn't seem to impact primary sensory areas, Corneocortical areas. And what we're thinking here is that it's probably because these are early maturing areas. So whatever if it is a disease process that we're looking at, it's targeting areas that mature later, And it's protec whatever processes that happen early are not interacting with this disease process, right? And what we know is that association cortices. continue to mature in your twenties when these sensory cortices are developed very early on, And synaptic pruning and myelination continues in these association cortices into the twenties, I think, although that kind of information is changing. so we're thinking that maybe the early maturing processes aren't so vulnerable to interactions with whatever pathology is causing schizophrenia, whereas these late maturing association cordices really are being targeted. Elvisha Dhamala: One aspect of your study that's really important is the transdiagnostic nature of your sample. So you included patients with a wide range of psychotic disorders, not just schizophrenia. How should we think about the generalizability of these findings across different clinical populations? and do you think the pattern might be related in some way to different diagnostic categories? Sid: Yeah, I think that's a really important point with this sample because it is a select sample in some ways, but in other ways, actually no, it's it's it's representative of what's in the literature. I think the way to think about it is that antipsychotics are the first line treatment for all psychosis spectrum disorders, regardless if you're coming in with substance induced psychosis, delusional disorder, psychosis not otherwise specified, schizophrenia, you you're given antipsychotics. So in that way, when trying to understand whether it's antipsychotics are causing iatrogenic harm to brain structure and brain tissue using MRI. I think it's a reasonable approach to be transdiagnostic, right? but obviously these people have different symptom profiles. So I think that you would find differences when you're dividing people up based on symptoms. I don't know if you'd find differences in DSM categories. I don't think these are really valid in terms of neurobiology. One of the things we see in this paper and this study, and I think a lot of people show this, is when you remove people from your analyses who have non-schizophrenia diagnoses, you do see stronger effects. And think we show that that you do see stronger thinning in the placebo group when you remove these other non-schizophrenia diagnoses. This is a common pattern, usually because the criteria required to diagnose schizophrenia results in stronger. severity of symptoms, So this is one of the reasons you see this effect, I think. I'm trying to remember the other part of your question, Lavisha. I think generalizability. Elvisha Dhamala: I think Sid: So the I will say one other thing with generalizability with this cohort, Is that if you look at the concert diagram, either in the clinical trial or one of our earlier studies, you'll see that actually it's only seven percent of people who came into contact with the clinical services at Origin where this trial took place that were illegible in the end or recruited, And the primary reason people exclude is that they've been exposed to antipsychotic medication before, So it was quite a select group of people, right? And it was to keep the experimental design clean, to really answer this I think neurobiological question in a clean way, that comes at the sacrifice of some generalizability, Because it's it's it's really a subsample. Because of the placebo, there was other safety considerations like people had to be living in stable housing, They had low risk of harm to sell for others. These are kind of again limiting categories in terms of the patient group. In saying that, when you look at the symptom severity of this group, it's comparable to other samples in the show. they had moderate to severe symptom severity, And you know, they're more severe than some of the other samples. Open samples like the human connect and project early psychosis sample. in terms of their symptom severity, as well as their functioning, it's actually comparable to a lot of different cohorts out there, although it is a sub subsample. And in the paper, and especially in that first paper looking at brain volume differences in this cohort in neuropsychopharmacology, we talk about generalizability in great detail because it's really important and nuanced nuanced point in this study, I think. Elvisha Dhamala: Another thing to consider is that your placebo group was also receiving psychosocial intervention throughout the trial. So how might that psychosocial treatment influence the thinning you saw in the placebo group and the absence of thinning in the antipsychotic group? Sid: So I guess with the psychosocial treatment that was being given to both groups, because it was being given to both groups, it's difficult to have any sort of mechanistic interpretation of the impact of that. The only one that I kind of land is that the thinning that we see when you don't give someone medication is over and above the psychosocial care, right? So regardless of psych psychosocial care, you're still seeing This thinning compared to healthy controls. That's kinda where we land on. I guess the absence of thinning in the antipsychotic group might mean that potentially there's an interaction between therapy and drug. We can't yeah, we because both groups got it, we can't really disentangle it, but perhaps, you know, the whole is more than the sum of its parts. I don't know if that's about saying right, but like in the antipsychotic group, potentially we didn't see any thinning, not because of the drug alone, but because there's an interaction between intensive psychosocial therapy and the drug. So that could be one other explanation of that. Look, we did have measures in this trial for how much people engaged with psychosocial therapy, so in that way you get a kind of a continuous measure of treatment exposure for the psychosocial therapy. That's something that we actually haven't looked at yet. Elvisha Dhamala: so my last question is if there's one key message you want clinicians and researchers to take away from this work, what would it be? Sid: I think that there has been this dogma in schizophrenia and psychosis research and clinical practice that antipsychotics cause brain changes for the negative, for the worse. And I think we show here that that needs to be reevaluated. This isn't the final word by any means, but at least that the doses that we were giving, for the length that we were giving them it was not the antipsychotics that are causing this thinning, And I think that we need to reevaluate that. No, a lot of people who still say this, A lot of prominent researchers as well as clinicians who take this as fact. So I think we need to reevaluate that. And this matters at the individual patient level, you know, I have talked to young people, I'm not a psychiatrist, a psychologist, and I see young people who are considering medication for various reasons and they often say things like, this will shrink my brain. I know this. I've read this, And I think that needs to be reevaluated. because it's definitely not fact, And this was the beauty of this study that it could really experimentally disentangle these things in a way cleanly to show that it's probably not the medications, at least the early stages that are causing these brain changes. So I think it has really important impacts for considering the risk benefit ratio when you're thinking about taking or when a loved one's thinking about taking these end these medications or when you think about prescribing them when you're thinking about brain health. I think that it has important implications for that. And I think it's a great way to show how imaging can have real clinical impact when done carefully. Elvisha Dhamala: That's all for today's episode. Thank you to Dr. Chopra for joining us to share his work on the effects of psychosis versus antipsychotics on the brain. For our listeners, you can find the full paper in Biological Psychiatry, and we'll include the link in our episode description. To learn more about our journals, please visit biologicalpsychiatryjournal.com. You can find The Biological Psychiatry Podcast wherever you get your podcasts. Subscribe now so you don't miss our next episode and follow us on our social media channels for highlights and updates. Thanks for listening. ⁓ This podcast is for informational purposes only and does not constitute medical or professional advice. ⁓ The views expressed are those of the participants and do not necessarily reflect those of the Society of Biological Psychiatry, ⁓ its family of journals, or its editors ⁓ All rights reserved, including those for text and data mining or use in AI systems. ⁓