Speaking of Science: Genetics of depression: risk and response to treatment with Dr Brittany Mitchell

In recognition of World Mental Health Day (10 October), NHMRC was joined by senior research officer at QIMR Berghofer, Dr Brittany Mitchell who is breaking ground on the genetics of depression to improve our understanding of risk, heterogeneity and treatment response.

By maximising her research through international collaborations, Dr Mitchell has been able to demonstrate a promising avenue to advance future treatments that mitigate poor health outcomes for people with depression.

Watch and listen as Dr Mitchell discusses the risk and response to treatment and the insights from the Australian Genetics of Depression Study.

Recorded on Tuesday 14 October 2025 at 11:00 am – 12:00 pm AEDT.

Video transcript

Professor Steve Wesselingh 0:00
This where we look into the stories of our nation's most outstanding researchers, and particularly researchers that we've supported, and on really important topics.

I thank all of you for joining us. I'd like to begin by acknowledging the traditional custodians of the lands on that we're all meeting on today across Australia, and acknowledge elders past and present, and certainly acknowledge Aboriginal and Torres Strait Islanders that are joining us today.

Before we get started, I'd like to encourage people to think about questions that you can put into the Zoom chat function and then we can put them to Brittany, who I'm going to talk about in a minute, but who you can see on screen. Also just emphasise that the webinars are recorded, and so you can go back and listen again, or if you have to leave, or you have people who haven't been able to join us today, they can go and have a look.

Professor Steve Wesselingh 1:07
Importantly, last Friday, the 10th of October, we acknowledged World Mental Health Day, and this day raises awareness of mental health issues around the world and mobilises efforts in support of mental health. I think all of us, I'm sure, have been touched by mental health issues, either personally or by a family member or a friend and it's certainly you know when you are and when you see it. We need to do more about this. This is such a difficult area. It's estimated that 1 in 7 Australians will experience depression in their lifetime, and this has the third highest burden of all diseases in Australia.

One of the statistics that I've got in front of me that makes me worry is that of those people who present with depression to our healthcare system, 30% will not respond to current management. That 7 out of 10 will get some benefit from our management, but 3 out of 10 with severe depression or anxiety will not get any benefit to what we currently have available. That really worries me and I think it is a significant gap in our ability to help people with mental health issues. It's a gap in our research, a gap in our understanding and a gap in new treatments coming forward, and that's why people like Brittany are so important.

Professor Steve Wesselingh 2:45
Brittany is one of Australia's inspiring researchers who are working in this area and thinking about better ways of managing mental health, looking at new treatments. She's an NHMRC Emerging Leadership Fellow and based at QIMR Berghofer Medical Research Institute in Brisbane, and her research focuses on the genetics of complex psychiatric disorders, with a particular interest in depression, anxiety and related traits. She combines large scale genomic data with advanced statistical methods to identify genetic risk factors in exploring sex differences and uncovering biological mechanisms underlying mental health conditions. Obviously, the better we understand it, the more likely we are going to develop treatments for these conditions. Again, I just highlight that we have a number of treatments, but a lot of those treatments are now 20 or 30 years old. We really have had very few new treatments in this area. I'm sure Brittany will tell us about this but it is an area where the research is so needed and I'm really keen to hear from Brittany.

Brittany has also played key roles in several national and international collaborations, including the Psychiatric Genomics Consortium, and is committed to translating those genetic discoveries into improved outcomes for individuals with mental health issues. Really privileged to hear from Brittany, one of our emerging leaders. I'm sure in 20 years time, Brittany will be one of our senior leaders in health and medical research, having solved all of these problems of mental health. Really looking forward to hearing from you, Brittany, thank you.

Dr Brittany Mitchell 4:39
Thank you for having me. I'm just going to share my screen and get my slides up. Are you able to see my slides?

Yes, okay, great. Thank you, Steve for the introduction, and thanks everyone for being here today. I'm really privileged to be able to speak about my research that I'm incredibly passionate about, as well as talk a little bit more about the Australian Genetics of Depression Study, which is the cohort that I'm going to base most of my research on, and tell you more about today.

Depression is the most common psychiatric disorder and the largest contributor to global disability. When we're looking at lifetime prevalence, as Steve said, it's approximately 1 in 7 people will suffer from depression in their lifetime. But these estimates vary across countries. There is this known sex difference in depression. Where females are between 2 and 4 times more likely to be diagnosed with depression in their lifetime than males. It has a mean age of onset around 26 years but more and more studies are showing that this is getting earlier. It's a recurrent disorder. Most people don't suffer from depression just once in their lifetime, with the average number of episodes averaging somewhere between 5 and 9 episodes in lifetime.

Dr Brittany Mitchell 6:07
In terms of annual prevalence, we're at about 5% of the global population suffer from depression each year. If we look at in terms of global disability and loss of productivity alone, it's estimated that it's costing the global economy about 1 trillion US dollars per year, and this is estimated to substantially rise by 2030 so it's clearly both a global health concern and also highly prevalent to Australia.

Part of what makes understanding depression and treating depression so difficult is its heterogeneity. Depression is classified by a variety of symptoms, and to get an official diagnosis, according to the DSM 5 of depression, you need 5 of 9 symptoms, of which at least one of these 2 that are in the center here.

But when you start to look at some of these symptoms that are classified as a single diagnostic symptom, you have things like insomnia and hypersomnia, or changes in weight or appetite, which could either be increased or decreased. It's highly possible that you can get 2 people that are both diagnosed with depression but don't have a single overlapping symptom. This makes it really difficult when we're trying to understand and better treat depression.

Dr Brittany Mitchell 7:27
But it's also highly heterogeneous in terms of its presentation. How it looks from one person to another, things like its chronic course, age of onset, comorbidity with other both mental and physical health disorders and then, as Steve was saying, response to treatment.

It's really commonly known that approximately a third of the people that are diagnosed with depression and prescribed treatment won't respond. That's a really high percentage. It's one of the highest percentages of all medical conditions, and that's really the focus of our work is to try and close this gap and either develop new drugs or better match people to medications that work for them sooner and so while I won't get into it today, there's a lot of debate around this heterogeneity, and a part of this complicated structure is that we're forcing something that's a syndrome into a specific diagnosis. There's some really nice reviews that I've listed there if you're interested to read about this more in terms of what we know about the genetics of depression.

Early twin studies suggested that the heritability was around 40% so this really means that genetics is contributing approximately 40% to the variation we see in the population when it comes to depression risk, so that's moderately high, and it's what we'd call a complex trait. This is where it's not influenced by a single gene or a couple of genes. It's influenced by many genetic variants, and each of these have a very small effect. It's the cumulative total or the predisposition that then influences genetic risk.

How we study complex traits is through mythical genome wide association studies, and I'll go through it a little bit more in the next slides. But what I wanted to point out is that unlike a lot of other medical conditions, the very early GWAS for depression were disappointing. The first large scale GWAS was published in 2013 and it had about 9,000 cases of people that had depression and comparing their genetics to about 9,000 controls of people that had never had depression in their lifetime, and they weren't able to identify any genetic variants associated with depression risk. This was really disappointing, because these sample sizes are obviously very large, it takes a lot of effort to get to this number. Unlike a lot of other conditions and even mental health conditions such as schizophrenia, for similar sample sizes, they were detecting risk variants. It became evident that for depression, we were going to need larger sample sizes, and that this was only ever going to be something that would be achievable through international collaboration and the pooling of worldwide genetic and clinical data.

Dr Brittany Mitchell 10:11
With this in mind, the Australian Genetics of Depression Study was launched in 2016 It was funded by the NHMRC, under the helm of Professor Nick Martin here at QIMR Berghofer. The goal of this was to collect a large cohort of Australians that had suffered from depression in their lifetime, as well as collect DNA and start contributing to the global effort to identify genes associated with depression, as well as trying to match people to, or understand the genetics of treatment response or why treatments are working for some people and not for others.

The recruitment strategy was to define the disease by the medication and so what this meant was a 2 pronged recruitment. The first was using prescription record data so Services Australia was able to mail out a letter to people that had had at least 5 antidepressant prescriptions in the last 5 years, and this letter directed volunteers to the study website that they could then choose to enrol if they were interested. At the same time, there was a nationwide social media campaign that asked for anyone to enrol if they had received medication for their depression. Once enrolled, there was a core module that asked about the experiences of depression as well as treatment, quite extensive information around what medications they had tried, what had worked, what had not worked, as well as a variety of satellite modules asking about other life experiences that help us entangle or do more nuanced studies into the relationship between depression and other factors. Then participants consented to providing a saliva sample that we could do DNA extraction as well as record linkage to the MBS and PBS data.

Dr Brittany Mitchell 11:58
Where we're at the moment, we have approximately 22,000 people enrolled in the AGDS, of which we've got genetic information for about 16,500 and the vast majority of these have also consented to allow us to link some of their MBs and PBS records.

Just a quick overview of what we've asked, so we've got this core module that asks all about depression and treatment, but we have all these satellite modules that ask about things like work and sleep or life events substance use and this has been really invaluable to us, being able to further the research, not only just into depression risk, but into other relationship and risk factors.

The last thing I wanted to point out about the AGDS is that very early on, we got indications that this cohort was enriched for people that were severely depressed or had been affected by severe episodes of depression in their lifetime. Some of these were for if we were plotting the age of onset. Our mean age of onset was between 15 and 19 years old, which is a lot earlier than the mean age of onset. I reported earlier across the general population and then when we were looking at a number of episodes. By far the majority of the cohort reported 13 or more episodes, which was our largest category that we had available in the questionnaire. Even though this is self reported data, this was given an indication that the people that had chosen to enrol in our study were affected severely by depression and so my research is around leveraging the AGDS to better understand three components, and this will kind of be the 3 aims of my talk going forward.

The first is using depression to understand risk and how genetics can predispose an individual to depression. The next is this heterogeneity, so why depression looks so different from one person to another, and whether genetics is part of this reason, and then the last component is treatment response and where we're at. It's very much an emerging field, but I'll hopefully look into how genetics can affect treatment response and why medications work for one person and not another.

Dr Brittany Mitchell 14:16
If we look at the first in genetic loci discovery, the premise is, if this is a pictogram of the world population, and we have somewhere between 15 and 20% of the world population affected by depression in their lifetime, and so the premise of studying a GWAS, if we say that this is now, sorry, I'm just going to put my laser around so I can, let's say this is the AGDS. We're then able to ask for a saliva sample, where individuals just put into a spit kit from which we can extract DNA and plate this DNA on these genotyping chips. These then get sent off to sequencing. What we're really looking for is these small changes in the DNA sequence and so in this example, there's a C versus a T in this position in the DNA, and then we compare these across people that have had depression versus a group of people that haven't. By doing that, we're able to then identify genetic variants that are associated significantly or a lot more common in people that have depression versus those that don't. Then this gives us a really nice starting point for 2 arms of research that tries to get a better treatment.

The first is novel biological insights and so it's quite commonly said that to be able to treat a disease, you need to understand it. Understanding the biology of what's causing depression allows us to make significant clinical advances in things of identifying therapeutic targets. This can be new targets for new drugs or new medications, or early biomarkers that could be used to diagnose depression and lead to its prevention. The other arm of research is more towards a personalised medicine. From these genetic risk variants that we've associated, we are then able to create individual genetic risk scores. For any person that we have their DNA, we can then calculate a personalised genetic risk score for depression based on what we know of the genetic risk variants. Then these can be then used in kind of a personalised medicine approach for diagnosis or therapeutic optimisation, so matching people with medications more likely to work on them based on their genetics.

Dr Brittany Mitchell 16:44
My research really is trying to tackle both of these arms and advance what we know in both aspects. In terms of low side discovery, we did a genome wide association study of depression in the AGDS. We had about 13,500 cases and 12,000 controls. We're able to identify one gene associated with depression in our sample. It's a gene called RBFOX1 which is the known depression gene and while it wasn't new or novel, we were really excited by the fact that we were coming up with signals that had already been associated with depression, and was a good proof of concept. This gene is involved in neuronal excitability as well as stress response, or hormone control in response to stress.

What we're really particularly excited about, though, is what we'd call a snip based heritability, and this is really just the proportion of the variance in risk that our genetic findings are explaining. We're getting about 15% which was substantially higher than any other studies at the time, where they were averaging at around 5 to 8% and so this was another indication, along with the kind of phenotypic measures of the early age of onset and the other forms of severity, and that our genetic cohort seems to be enriched for severe depression, and that the genetics are picking up some level of severity in depression risk. We combine that with the largest known GWAS of the time for depression, and by adding in our sample, we were able to identify 23 new genes for depression. We were really excited about this. This was on a global scale, and there's a lot more work to do to really understand these and see if these can be new drug targets or help to us to understand the biology of depression more. But we were really proud of this contribution.

Alongside that, the psychiatric genomic consortium was undergoing a much larger GWAS meta analysis at the time, and so we were able to contribute our data to that. This study came out this year. It was published in cell earlier this year, and through that, we've identified almost 700 genetic associations of depression. That's a really mammoth increase, and that came from an incredibly large sample size. Overall, it had about 680,000 cases of people that had had depression, and comparing their DNA to over 4 million people that had never had depression. This was a real mammoth worldwide effort. The paper has 375 authors from 29 countries, and I'm really proud of this work.

Dr Brittany Mitchell 19:35
The AGDS contribution was substantial, and we were really proud to be able to lead the Australian arm in the world's biggest study of the genetics of depression. But I also think it's a really nice example to show how much advancement can be made when people come together and they're willing to work together, share data and share expertise for the greater good of kind of moving the needle of what we know when it comes to genetic predisposition for depression.

This is where we're at. This is the very latest in terms of overall genetics for depression risk. The other arm is that personalised medicine arm. In the AGDS, we were able to calculate personalised genetic risk scores for everyone in the sample, and when plotting that, we showed that it was significantly able to predict depression. What this figure shows is that individuals that are in the top 10% of highest depression risk were about 6 and a half times more likely to be diagnosed with depression than individuals that were in the lowest decile, and so that's quite a big stratification, and already shows that just using genetics alone, if you take the tail ends of those distributions, that there is a significant difference in risk based on on genetics.

What we're also able to show is that our genetic risk scores also map to severity. It wasn't just overall risk, but we're able to show that people with higher genetic risk scores were more likely to have formal diagnosis of depression, experience more depression symptoms than less, and have recurrent episodes versus just single episode of depression. This was also nice to see in that the genetic risk scores we'd calculated mapped to severity measures and not just overall risk. If we compare that to the results of the PGC from this year, on average across the 48 samples that contributed to this big meta analysis, they had the same really nice stratification across the deciles, where individuals in the top 10% were about 5 times more likely to be diagnosed with depression than those in the bottom 10%.

Dr Brittany Mitchell 21:51
Then they did this great kind of separation, where they separated cohorts that were from clinical backgrounds being formally diagnosed by clinicians, versus what they'd call community cohorts. These are larger scale kind of biobanks, things like the UK Biobank, and they showed that there was this clear separation, at least in the upper deciles, meaning that individuals that had higher genetic risk scores were more likely to come from a clinical cohort. It was really this matching of genetic risk to or higher genetic risk to another form of severity. That was really nice to see, but also interesting to see that the AGDS in itself was higher than both of these estimates. Again, showing that we've got quite a severe sample.

Something our lab's particularly interested is the sex difference. This 2:1 ratio between females and males. This is the same pictogram, just showing on average, that 2:1 ratio. We know there's this known prevalence difference, that prevalence difference persists across many different countries and cultures, but there's also differences in presentations, so females that suffer from depression are more likely to have a subtype called atypical depression. It's characterised by symptoms such as weight gain, increased appetite, hypersomnia, and has been linked to forms of immunometabolic depression that involving both inflammatory and metabolic pathways. Males that suffer from depression are more likely to exhibit externalising behaviour. This is things like aggression or risk taking behaviour, are more likely to engage in substance use, and so we're really interested to understand this more. There's many potential explanations, and it's more likely that all of these play a role.

Things around reporting and help seeking behaviour. We know that there's a there's a known difference, and there's a known under-reporting of males when it comes to seeking help particularly for mental health disorders, but it does persist across all medical disorders. That 2:1 difference might be impacted at least in part by males being under reported or not diagnosed, even though they were suffering from depression. Environmental exposure, we know there's known many sex differences in terms of the environment exposure, things like exposure to trauma, caregiving responsibilities, but also things like war, PTSD, job related stress, and then there's biological. We know there's biological differences between males and females, but there's been very little research in whether there's biological differences in genetic risk between males and females. That's what we wanted to look at.

Dr Brittany Mitchell 24:45
We did a study led by my postdoc, Dr Jodie Thomas, and she was able to find some specific genes that were associated with depression in females versus males. Obviously, these numbers are a lot smaller than what you see the 600 I've just talked about in the overall genomic consortium, but this was because we're dealing with much smaller sample sizes. We had 5 international cohorts, but the study really led out of Australia and led out from the AGDS. But we're really excited by some of the results.

One of them this particular gene, it was associated with depression only in males and not in females. It's a gene that's never been associated with depression before, so we're really excited about that. It's located on the X chromosome, and so that's really interesting. We know obviously the X chromosome is very strongly linked to biological sex, and it looks like this gene is playing a role in some kind of regulation of immune response or inflammatory pathways, and so we want to look into that a little bit more in future studies.

But the other thing I wanted to highlight from the study was the overall looking at genetic architecture. We calculated the snip based heritability, so the variation is explained by our genetics in depression risk between males and females when we looked at them separately, and we found that it was significantly higher in females. This is really showing that genetics is playing a larger role in depression risk in females than genetics is playing in males. To kind of look at that from another way, we used a different method where we estimated the total number of genetic variants that contribute to depression between males and females, and what we found was that there were about 7,000 estimated genetic variants that contribute to depression in both sexes, but that females had an additional 6,000 so there's close to 50% more genetic variants contributing to depression, and there were no sex specific male genetic variants. This is just another way of showing that from all our analyses, it seems like genetics is playing a stronger role in depression risk in females than it is in males.

Dr Brittany Mitchell 27:06
The next part of my talk that I want to talk about is heterogeneity. This is looking at the genetic contribution of why depression looks so different from one person to another. The second part of the sex depression differences study was trying to understand those differences between metabolic and substance use, between males and females, and what really stood out as significant when we were looking at genetic correlations, was there was a significant difference between females with depression and cardio or metabolic traits. Things like BMI, waist to hip ratio and metabolic syndrome, and so those were significantly higher correlated with female depression than they were with males.

Then similar, the mixer analysis, which was similar to that I've just presented, showed that there was about 8,000 genetic variants shared between female depression and female BMI, compared to about only 4,000 between male depression and male BMI. This indicated that there are shared biology predisposing females to higher depression and also to higher metabolic rates, such as BMI, than there are to males. This is really nice, because this might be a validation of why we see these symptoms come up a lot more in females when they're suffering from depression. This paper was just published pretty recently in August, so if you're interested, please take a look. But what we're most proud of is that we've made everything publicly available, all the scripts, and then the data and our results are there to be downloaded. We're very eager for people to use our results. I hope this is the start of showing that we do need to look at sex when we're looking at genetic studies of mental health disorders, and that there's a lot more to be done. We're keen for people to use our results and dig into this a lot more.

The next part of heterogeneity is around subtypes. There are many different subtypes of depression, and in my Biological Psychiatry paper that I talked about earlier, part of what we did is divide the AGDS into different subtypes. We had things like anxious depression, atypical depression, depression with psychotic features, etc. I wanted to look at the polygenic risk score profile, so not only just the risk for depression, which I've talked about, but does your genetic risk for other mental health disorders or related traits start to separate these different types of depression.

Dr Brittany Mitchell 29:43
There's a very complicated figure that I won't spend too much time on, but we did show that there were these differential patterns of association. Things like schizophrenia, showing that it was a genetic risk for schizophrenia was associated more likely to have anxious depression or depression or psychotic tendencies and less likely to have atypical depression. The take home from this was that we can start to get these differential patterns of association when we start looking at genetic risk across a variety of disorders, and that this might help us understand depression more and get more accurate diagnosis if we look beyond just a single polygenic risk score, but stop building these profiles and look at these different patterns of association.

We work very closely with the Brain and Mind Institute on the AGDS, and Dr Jacob Krause has done a lot of really nice work using the AGDS. I just wanted to highlight this study, which was challenging the notion that there's situational or environmental depression versus biological depression, and that these 2 things are completely separate. What he showed was that overall, when we're looking at the number of stressful life events, and so these are things different types of trauma that individuals have reported in the AGDS, there were significant associations with your genetic predisposition, predisposition. This were things across a variety of polygenic risk scores, so things like ADHD, but also depression risk was associated with how many stressful life events you reported having, and that, again, there were some of these differential associations when we start looking at the types of stressful life events, things from natural disaster to different types of trauma or witnessing injury or death. Again, just challenging this notion that you can separate the environmental effects from the biological effects, and that they are very intertwined, and that there's probably a lot of interaction going on between the 2.

Then the last kind of thing I wanted to talk about was around seasonal affective disorder, or the effect of season, circadian rhythms, light exposure. This is something we're really interested in. It's been driven primarily through the Brain and Mind Institute, but it's well known that light has profound influences on neurobiology, and that exposure to light activates all these areas in the brain, and that these are the similar areas that also play a role in mental health. We've also shown that people with depression get very little daylight exposure compared to people that don't have depression. That's both shown on free days and work days and it does make sense. There's obviously a lot of reasons, but you can understand that people that are feeling really depressed are probably less likely to go outside. We've done a variety of different studies in this area and so we had a student come visit us from Amsterdam, and he looked at genetic predictors of seasonal affective disorder, and was able to show the chronotype, which is whether you're a morning person or an evening person, was significantly interacting with your latitude of where you lived in Australia to predict seasonal affective disorder, or even how much the change of seasons was able to affect your mood and your mental health.

Dr Brittany Mitchell 33:17
This is a real, really novel study that we're quite excited about, and that it's showing there's very few studies looking at the effective season in the southern hemisphere. Obviously, our seasons are not as different or distinct from one another as perhaps in the northern hemisphere. But what we're able to show is that even in Australia, where you live, can interact with your lifestyle, and whether you're a morning person or an evening person, or evening person, to then have an effect on on the effect of season influencing your depression or your mental health. We're excited to dig into this a little bit more and have a look at how much of this is season versus light exposure at different times of the day, or potentially even temperature. As we know, temperatures vary quite substantially as we move across Australia.

Jake's also done a study looking at response to treatment, and showed that people that are evening people, so like night owls, are less likely to find the most common antidepressants or SSRIs or SNRIs to be effective, and that night owls are also more likely to have side effects as a result of their medication. This is really interesting, and pointing at the circadian influence on depression risk, and something we want to look at more.

Dr Brittany Mitchell 34:36
Going forward, Jake's leading this study where we've got these new wearable sensors that are able to examine body clocks. We re-approached members of the AGDS to enroll in the study. They can use these little pins, and these pins are able to monitor circadian rhythms as well as light exposure at different times in the day and hopefully move us towards a place where we can start using like based recommendations on what to do for your mental health, and hopefully incorporate that with genetics to move towards a more personaliSed treatment protocols.

Okay, the last part of my talk I wanted to talk about was treatment responses is very much an emerging field, and so this is really the focus of my fellowship is to try advance what we know in treatment response and the genetics of treatment response, but it's really a difficult area. One of the big questions is, how do you define response? Obviously, many studies have assessed this in many different ways. Ideally, you'd be looking at kind of a clinical study where you've got a clinician rating depressive symptoms before, during and after treatment. But as I spoke about earlier, genetic studies, you also need very large sample sizes, and so that's really difficult to get in that kind of setting. A lot of the biggest studies have asked self reported questionnaires, which is a start, but that's also retrospective, and we know that retrospective reporting, particularly when it comes to treatment response, is not always particularly reliable.

Dr Brittany Mitchell 36:14
Then the other complication with depression specifically, is that treatment is so variable, and there are many different antidepressants, many different classes of drugs. There's a lot of prescriber preference going on, and so that people are prescribed antidepressants based on who's prescribing them, and not necessarily on specific symptoms and there's a lot of variability, which makes it really difficult to study in AGDS, we have a retrospective question, asking, how well did a particular antidepressant work for them? Based on that, we have tried to define responders and non responders. We did a GWAS of that in AGDS, we found one genetic variant, but it was lying in a section of genome where it mapped to a gene that was really difficult to kind of - there's not a lot known about the gene or what it might be doing. We've contributed these results to the world's biggest GWAS of treatment responses being led out of Norway. What they've shown, this is a paper that's in press at the moment so should hopefully be out soon, this is just an overlapped, overlaid Manhattan plot. It's a little bit difficult to see, but what they've done is they've done a GWAS for SSRI response, and found 2 genes associated with SSRI response, one gene with SNRI response, and then 2 that were with either type of antidepressant response. These are the first kind of replicable and genetic variants, and we're busy at the moment looking into their biological pathways and whether they give us new potential drug targets, or that we can repurpose old medication to work on.

These medications or already approved medications target these pathways to treat depression. But what we're also able to show is that if we created a genetic risk score from these results, we were able to predict non response across 5 different cohorts. They weren't all significant, but in the AGDS, it was treat non response and so this was really encouraging in that even though we only have 2 genetic variants identified for SSRI non response at this point, if we create a genetic risk score from that within our sample, we were able to predict who was not going to respond to SSRIs. That's really encouraging. That's a really nice positive step forward towards this personalised medicine, where hopefully we can start incorporating genetic risk scores into modelling frameworks to predict who will respond to different types of medication, and reduce that timeframe and the 30% that don't find medication or treatments that work for them.

Dr Brittany Mitchell 39:03
The other way we've tried to look at defining responses using prescription record data. In the AGDS, we have a subset of people that allowed us to link to their PBS data, and so I've been doing a lot of work on that recently, and I think anyone that's worked with prescription record data will know that it's really messy and really difficult data to work with. This is just a snapshot where each line is a different person over about a 5 year period, and each is a prescription of an antidepressant that they filled. You can see that there's a lot of, it's very difficult to discern patterns and there's a lot of changing of medication, but when we impose different rules, you can see that you do find individuals that meet different criteria.

At the top here, I've shown these are people that have been on the same antidepressant and for a consistent period of time. You could infer from that, that the medication is working for these people and that they're able to stay on the same medication. In contrast, you also get people that are on many different medications that are changing consistently. There's overlapping medications, indicating that they're either changing regimens or augmenting a certain medication with another to try have a better effect and so you could also potentially discern from this medication they're more, you know, complex treatment or treatment resistant. We've done a little bit of work on this.

A PhD student in Oxford called Alicia Walker, who's done a little bit of looking at genetic risk scores predicting these antidepressant behaviours, found that some genetic risks, so things like your genetic risk for depression, ADHD, bipolar disorder and neuroticism, were significantly associated with how many different types of antidepressants you're taking, as well as the different class diversity. If we take that as a proxy of kind of non response or complex treatment, showing that these people that have higher genetic risk across these 4 domains might be less likely to respond to medication than people that have lower genetic risk.

Dr Brittany Mitchell 41:24
I then wanted to take this forward and try to define groups of people. What we did is we stratified our individuals into 7 groups, 2 SSRI responders, these were people that consistently took SSRIs across the time period, 2 groups of SNRI responders, and then 3 non response groups. I ended up defining non response based on having electroconvulsive therapy. It's a very extreme form of treatment and so anyone that indicated that they were either recommended or had undergone ECT therapy, we classified as non response people that were later diagnosed with bipolar disorder. We have a group of people that originally were diagnosed with depression, underwent a variety of treatments that didn't work, and then were later diagnosed with bipolar disorder.

Then our last group was the augmentation group, where individuals were taking an antidepressant, but augmenting this with either lithium or an antipsychotic, indicating just a more complex treatment regimen than just being on a single antidepressant to treat their depression. If we have a look at these 7 groups, what we were able to show is that those in our difficult to treat, or our non response 3 categories at the end here were reported a higher number of side effects from medication than those in our response categories.

There's a lot of research and public interest in metabolising genes. There's genetic kits and companies that are recommending you have your the cytochrome genes tested, and that is really just a liver enzyme that controls how you metabolise drugs, and that this might influence how effective a drug is or not. We had a look at the 2 most common ones across our groups, but we didn't find any significant differences in the metaboliser profiles. There's definitely a lot more work to be done to really validate how reliable these cytochrome genes are in predicting drug response.

Dr Brittany Mitchell 43:29
Then our preliminary work was again looking across genetic risk profiles and showed that we were getting some association specific to groups, so things like a high genetic risk of bipolar disorder, whether or not they actually received a diagnosis of bipolar, but it's your genetic risk for bipolar, it's more likely to mean that antidepressants are not responding or you're less likely to respond to antidepressants. We had a negative association with your genetic risk of BMI, so individuals that were predisposed to having a lower BMI were more likely to stick with a regimen of sertraline or the most common SSRI response. This is really interesting, a really or probably the most common side effect, or one of the most common side effects of sertraline, is weight gain. It is plausible that people that have a lower genetic risk of having a lower BMI, more likely to have a lower BMI might see a response from this medication or be less deterred by the side effect of weight gain. This is really preliminary work. We're busy categorising these groups and seeing what we can find.

But the exciting part of this research is the new offshoot, which is the AGDS Salomic study. It's been headed by Professor Naomi Ray That's at UQ and University of Oxford, and so this is where we're setting up a cell line resource. We've gone back to these people in the 7 groups asked for a blood donation and from this we're able to cerive stem cells and eventually organoids, which will mean we'll have these cell based assays of people where medications work for them versus people that haven't had antidepressant medication work for them. This opens up a really exciting, like cutting edge avenue of research, where we can start testing exposure if we put different drugs in there, see how the biological differences or the effects that those have on these stem cells or mini brains, as well as characterise across a variety of different biological assays and produce, potentially, you could see in the future, potentially, a blood test that could then tell you which antidepressant works best for you, and so that's the ultimate goal. That's where we're hoping to head to. We're really excited about this research going forward and what we might be able to do from there.

In conclusion, it's clear that meaningful progress in psychiatric genetics requires worldwide collaboration, but as our sample sizes are growing, our phenotyping and making sure we're measuring the right thing is becoming a greater consideration. We feel like the AGDS is a really nice proof of principle study demonstrating the utility of online recruitment, where we've been able to maximize the sample size, but also have this really great phenotyping and understanding the genetic influences on depression risk, its variability and its response to treatment, is a really promising avenue to advance our treatments or mitigate poor treatment outcomes.

Dr Brittany Mitchell 46:35
But it's never going to be the be all and end all and so what I'm really excited about is future offshoots that we can do with the study, where we integrate what we know with genetics with other epidemiological or other types of phenotyping. There's some incredible research being done in like the field of digital phenotyping, and so I'd be really keen to have a look if we can contribute or combine our genetic risk predictors with what we can get from other types of phenotyping to build these models that are accurate and able to better match people with more effective treatments.

But I hope I've shown a little bit today of how genetics is reshaping our understanding of depression, from uncovering the biology to how we predict and treat it. Lastly, we're super eager to collaborate, so please get in touch if you think any of our data could be of use to you, or if there's any work we could do together in the future in this space.

I'd just like to thank you all for coming and listening to me. The chief investigators of the AGDS study, as well as the analysts whose work I've presented today, and I'm happy to take any questions. Thank you.

Professor Steve Wesselingh 47:52
Thanks very much, Brittany, that was terrific. So good and really a lot of information about a really difficult topic and some enormous international collaborations which are really, really impressive. Maybe we'll get you to stop sharing your slides great.

We've already got some questions online, which I haven't actually had chance to read yet because they look quite long. I'm going to give you a question, and then I can start looking at the questions online.

I was really impressed with the international collaboration and the Cell paper that came out of that. That was enormous. Would you like to comment to the people online about how was that difficult? How did you go about that? You know, just an enormous collaboration. Maybe just a bit of commentary on the international nature of the research, and really that you went from what you were finding in Australia, which was important, but when you combined it, it was just so much stronger. Maybe comment on that.

Dr Brittany Mitchell 49:02
Yeah, I think it's become very obvious very quickly that no cohort, or even no country, is going to really be able to achieve the sample size as necessary to identify the full component of the genetic contribution to depression and so that's really where the PGC has come in. It's really great model of data sharing, and it's this open science model where everything's transparent. Anyone that wants to be involved can be involved. There's no application process, and it's just a it's a really great environment where everyone's very welcomed. It's a group of scientists around the world that knows that to advance this field that we're really passionate about, we do need to share our data and be very open to collaborating.

While the organisation of that paper was very intense, we had quite a big analyst group of about 20 of us that all divided different components of the research to get it done. It did take a long time. It took about 4 years to come together. But we're very proud, and we're very proud, particularly of the Australian contribution, because it was really significant to that paper.

Professor Steve Wesselingh 50:16
Thanks for that. I'm just quickly going through the questions online, but a number of the first ones are really going to the genetics, and whether you're finding something new in the genetics and some mutations or some genes that are pushing you towards I mean, I guess I would summarize that our past work has all been really around serotonin and dopamine. Are there new clues, and you know, those drugs do provide benefit, but 30% don't benefit, and even those who don't benefit a lot, always.

One of those questions links it to possible issues of infection, but are there some new pathways that we haven't thought about that are coming out of the genetics?

Dr Brittany Mitchell 51:06
Yeah, there definitely is. I think the difficulty has been in that we've jumped so quickly to this very large number of genetic variants that there's a lot of new pathways, and to understand or to unravel them from one another has been very difficult, but what the 2 that stood out in the Cell paper that we're quite excited about, we were able to identify these pathways and then look for already approved drugs on the market that were treating different disorders, but that worked on that same pathway.

The one of the 2 that's coming to mind at the moment is methaphenols used to treat narcolepsy and is quite often prescribed to shift workers to keep them awake when they need to be. Their circadian rhythms are disrupted quite a lot, and we found that that could potentially be a new treatment for depression, or that it was acting on these new biological pathways, which makes sense when sleep and circadian rhythm is such an important part of depression. We're really excited around that, but we've definitely got a very long way to go to look at the function of those pathways and what they're doing.

Professor Steve Wesselingh 52:15
You said 2 pathways? So the narcolepsy receptor was one pathway. What was the other?

Dr Brittany Mitchell 52:20
Yeah, I believe it was chronic pain. It was a drug that I can't remember offhand, but it was a drug that was used to treat chronic pain and inflammation, and so that coming up as another important aspect. We know inflammation plays a big role in depression. There's a lot of connection with gut health and things like IBS and depression risk, and so it was a drug that I can't remember his name at the moment, but used to treat chronic pain.

Professor Steve Wesselingh 52:48
Alright, one of the other questions online was about insurance, and so is your pharmacogenomics advanced enough to impact on issues like insurance and other aspects. I'm sure you've thought about that.

Dr Brittany Mitchell 53:07
Really good question, and we're not at the point of being able to take the genetic risk scores into the clinic, yet. We're scientifically able to predict but they do need to be stronger to be really useful in the clinic. But a big part of the reason why we're also not ready for that is that we're advancing the science arm of the research, but at the same time, there needs to be legislative, ethical advancements in the controlling and sharing of that data. We've all seen, there's been stories of 23andme and all the things that can go wrong when genetic data is leaked or shared with the wrong people. It's a really big problem, and it's going to need really strict laws and enforcement on how it can be used and how it can't be used to protect the consumers.

Professor Steve Wesselingh 53:58
Yeah, that was a really important question, but I actually misinterpreted the question. The question is actually about whether health insurers should be funding some of the genetic and pharmacogenetic testing for individual patients.

Dr Brittany Mitchell 54:14
It's a great question. I think it would be great if they would, because we need more funding desperately. There's also a risk, when you start getting companies with private interests involved in research. Obviously, there needs to be a lot of regulation so that there's no bias of results, or that it is for the science purposes and not to advance company goals. But it's definitely an area we need more funding, and so the more the better.

Professor Steve Wesselingh 54:47
I have to say, all of these questions start with great work, Brittany, such a great talk. So brilliant.

One of the later ones talks about proteomics and epigenome. Have you thought about linking your genomics with proteomics and epigenomics, you know, sort of multiomics approach?

Dr Brittany Mitchell 55:11
Absolutely. It's the future of where we need to go. The thing in AGDS is, so we collected from saliva samples, where you can't really measure that. You need to take it from blood samples, which are obviously more difficult, but also a lot more expensive to collect on scale. That's where we're excited about the senomics antidepressant response study, because we've got blood from them. We are going to be able to get all these different layers of different omic data and then start looking at like an assay across them, a predictive assay that can then be built from blood samples,

Professor Steve Wesselingh 55:46
Then there's a question on naltrexone being trialled in ME/CFS or long covid, and whether there'd been any overlap with depression, particularly with the sort of chronic fatigue syndrome, type syndromes.

Dr Brittany Mitchell 56:04
It's a great question. I don't know off the top of my head. There's definitely been a lot of overlap, even with response to covid or susceptibility to covid infection and long covid infection. We know that there's definitely immune related pathways predisposing both, and that there probably would be a lot to look into there, but we just haven't got there yet.

Professor Steve Wesselingh 56:35
Alright, we are continuing to get in questions. We're running out of time. But there's one about autism and overlap between autism and depression. Have you seen anything in your work in regard to that?

Dr Brittany Mitchell 56:46
Absolutely, yeah, there is an overlap. There's an overlap between depression and all mental health disorders, anxiety being by far the largest or the strongest overlap. But really trying to unravel them is one of the next. What I see happening in the next decade is separating what is specific to depression versus all the other disorders.

Professor Steve Wesselingh 57:09
There's a question about the cello mix and whether there would be another call for blood from people who might like to donate.

Dr Brittany Mitchell 57:19
I'll have to chat to Naomi, who's leading that. I'm not sure, but obviously, the more people we have, the better the study would be, and that would be really great.

Professor Steve Wesselingh 57:28
Perhaps the last question from me. Lithium has been a drug that's been around for a long time, was discovered by an Australian as a treatment for depression. Has any of your genetics or any of your work indicated or helped us understand how lithium actually works, or given more hints about the use of lithium in the treatment of depression?

Dr Brittany Mitchell 57:51
No, it's a great question. Lithium is a drug. It's most commonly prescribed to treat bipolar disorder these days, and there's been a lot of fantastic work in advancing the response to lithium. Just because for people with bipolar disorder, they, more often than not, are on lithium, so they don't have that very wide array of different types of medications that we have in depression.

We have seen that it has been effective for people that have an antidepressant with lithium as an augmented thing. But I haven't looked too much into the results of the other reason, there is definitely research out there, I just haven't been particularly involved in that on what lithium is targeting in the brain and how it's working.

Professor Steve Wesselingh 58:44
Time is up, but I'd just like to thank you so much for a terrific talk, great work, and fantastic international collaboration, just brilliant and on such an important topic. Because, as I said at the beginning, I think we all know people who are suffering from depression and aren't necessarily responding to current treatments, and we need a lot more work and your work is just so good in that area. Thank you very much. Great talk and great work. Thanks a lot.

Dr Brittany Mitchell 59:15
Thank you very much. Thanks for having me. Bye.

End of transcript.