SAiGENCI Public Lecture

Added: 2026-05-14

[00:01:12]Uh I'd just like to share a few words uh for acknowledgement of country.

[00:01:27]Okay. So Adelaide University respectfully acknowledges the Ghana, Boandic and Bangala First Nations people and their elders, past, present, um, who are the first nations traditional owners of the lands that are home to our campuses across South Australia. We also acknowledge other First Nations lands across Australia with which we conduct business, their elders, ancestors, cultures, and heritage.

[00:01:53]So just briefly a couple of housekeeping items. Uh so toilets, ladies toilets are located on this level uh just across from the theater whilst gent's toilets are one floor down accessible via the lifts or the stairs.

[00:02:14]So for those who are attending the lecture for the first time uh the South Australian immunogenomics cancer institute or sigensi is the newest cancer research institute in South Australia.

[00:02:27]It's a collaboration between Adelaide University and the central Adelaide local health network or kalen. Our goal is simple but powerful to have fewer patients develop cancer and to improve the lives of those with cancer. So now in its fourth year uh we hold this public lecture annually to share our research and to interact with the wider community. So just a big thank you again uh for your interest and engagement by attending tonight.

[00:02:57]So without further ado, uh on to the main event. Uh it's my pleasure to introduce our two speakers for tonight.

[00:03:04]Our first speaker is Professor Lisa Butler. She is a program lead in the resistance prevention program at um Sency and a group leader of the prostate cancer research group. She's a world leading expert in prostate cancer research and she holds key uh executive positions in the Australian prostate cancer bio resource, the prostate cancer foundation of Australia and the Australian and New Zealand Eurogenital and um prostate cancer trials group.

[00:03:35]Our second speaker is Professor Lan Nuin. He's the program lead of the computational systems oncology program and group leader of the integrated network modeling lab at Senci. He's also an ARC future fellow and he's head of the integrative network modeling lab at the mache biio medicine discovery institute. Professor Nuin has extensive experience in integrating advanced computational uh modeling linked up with cutting edge biological techniques to understand and combat cancer. So without further ado, I'd like to welcome uh our first speaker, Professor Lisa Butler to the stage.

[00:04:26]Thank you very much Michael and good evening everyone. It's such a pleasure to be here tonight to talk to you about some of the work that we're doing in Sigency mainly focused on trying to understand and prevent cancers be from becoming resistant to therapy. Now you might wonder why is that so important?

[00:04:44]Well, cancer obviously affects so many of us um with a million or so of Australia's population diagnosed over the last decade and 170,000 approximate new cases every year. Now, while we're making great strides in uh improving survival from cancer, the fact remains that resistance to cancer treatment is the key driver of mortality and cancer and something we desperately need to overcome.

[00:05:13]Now our program at Sugency is of six very different groups but all working on this similar goal and our focus is very much on the two most uh commonly diagnosed cancers in Australian men and women and that is prostate and breast cancer. And in fact some of you may not know that prostate cancer now has the dubious honor of now being the uh the most commonly diagnosed cancer overall in Australia. So it's a huge health problem.

[00:05:42]Having said that, as I mentioned, we've made some huge strides over the last 20 years. And this is the latest data uh from the Australian government. Uh if you look at the mortality rates from breast and prostate cancer, this is just prostate cancer, but breast is extremely similar. We've seen a very gratifying decrease in mortality over the last 20 years. And that largely reflects factors such as better and earlier detection, um new treatments being developed. uh a lot of research has really gone into this um which is great but the problem of course is that our population continues to grow and our our population is aging. So despite this impressive decrease in mortality to the point that both of these cancers have a greater than 90% survival rate now after diagnosis overall uh the total number of deaths from uh prostate and breast cancer continues to rise to the extent that in South Australia for example almost a man dies every day from prostate cancer. So it's a huge problem and there's an even much greater number of of uh people who are living with either active cancer or a diagnosis and that uncertainty about what's going to be the progression of their disease uh in the longer term. So we've made some improvements but we need to do more. So our next challenge is really trying to stop cancer from adapting and starting to become resistant to therapy so we can achieve much better longer term control of these cancers.

[00:07:16]So our program is very much focused on very molecularbased research uh that is trying to understand the mechanisms by which cancer cells can adapt to become resistant to therapy and we really focus very much on individual patients. So we know that just like every person is different, every cancer is different as well. And so each might have a different way that it's evolved to become resistant to treatment and therefore would need a different approach uh to to treatment in the future. Um so our goal is really to to use this the learnings that we uh that we uncover to then as a basis to design and hopefully move towards the clinic uh new therapies for patients that can uh eventually uh ultimately delay or prevent that that progression to uh to advanced uh treatment resistant cancer.

[00:08:09]So when we think about cancer resistance to therapy, it's often described in a very simplistic way and we we think of it this way ourselves at times. Uh and often it's kind of compared to antibiotic resistance in the sense that there's innate and acquired versions of uh of resistance. So for example, when we think about innate resistance, um sorry, I don't know if you can see my my marker there. Um uh what we're talking about is the fact that you know cancers are often made up of very very different types of cells each with different genetic lesions and other modifications that mean that they're all slightly different in their sensitivity to a treatment. So when you treat on the right that sort of tumor, you're going to kill the cancer the sensitive cancer cells, but the resistance ones are going to remain and they're ultimately going to expand and take over the tumor. On the other hand, we know that there's also a certain amount of acquired resistance. Now, what that means is that cancers can be initially sensitive to therapy, but during that process, they acquire genetic, epigenetic, and other types of changes that allow them to suddenly survive and uh and evolve to become a very treatment resistant cancer cell. And I think in in all reality, it's a combination of these mechanisms.

[00:09:31]And you'll hear more from professor Nuin about, you know, more sophisticated ways that we can sort of think about how this evolution occurs.

[00:09:40]But the fact remains if we don't understand how this evolution process occurs, we'll never really be able to intervene or predict when that's going to happen. So in our program, we have uh six different group leaders, all with very different expertise. So we're all applying our various uh expertise uh whether it be in genetics, in epigenetics, in metabolism, in cell stress to try and understand this evolution of cancer cells to a treatment resistant state. And then in a way to take our discoveries from the laboratory to the clinic, we have group leaders such as Michael Roy who just uh introduced me who's a medicinal chemist and he has some really innovative ways to develop agents that can either act as drugs or for targets that we find that might not be amendable to drugs, ways to degrade and attack drivers of this treatment resistance. And of course then we want to move it closer to the patient uh in terms of clinical translation and we're very fortunate to have Chris Sweeney the director of our institute as one of our group leaders. Now he's a molecular sorry not he is a molecular biologist as well but he's also a medical oncologist and so he's very experienced in clinical trials for cancer and so he really has helped us in terms of devising a pathway from the lab to the clinic but also as an active research scientist as well. his clinical perspective is helping us in designing and um and analyzing and interpreting uh our more biological findings.

[00:11:16]So with that overview, I thought I would just give you an example of some of the work that my group has been doing uh to tackle this problem and my group uh focuses on prostate cancer. Now the last 20 years or so of prostate cancer clinical management has really been a bit of a golden age actually because prior to that so if we take a step back in time to 2005 you can probably see from this uh this schematic here that the options were pretty limited and you can sort of see that they're very much a one-sizefits-all category. So in that in that time when uh if men were diagnosed in an early stage with treatment sensitive disease there was the option for curative interventions such as surgery or radiation or for very low risk cancers for active surveillance and watching. uh if there was a recurrence after this uh localized therapy, the main option was hormonal therapy and most people respond really well initially to hormonal therapy in terms of their cancer burden. However, uh there is inevitably resistance that occurs and then it progresses ultimately to a an aggressive treatment resistant phase of disease that's called castration resistant disease. And you can see that the uh options here were pretty limited either observation chemotherapy that wasn't particularly effective or paliotative care. Now the picture in 2025 is very different. So now you can probably see from this uh this particular diagram that now we have many many more agents and that's been the result of huge investments in terms of funding uh and in terms of research and key clinical trials that I've highlighted here in blue that have shown that a number of new agents some are hormonal some are not uh but essentially these are agents that have been shown to improve the outcomes of patients from prostate cancer. So it suddenly goes to a situation where it can be a bit more personalized. Patients have many more options that they can choose from and the uh the lifespans are extending. So that's really great news because that buys time that more research can happen and more options for people to try.

[00:13:34]However, the main central problem remains in the sense that all of these agents while they do uh extend lifespan in patients, none of them are curative.

[00:13:45]So ultimately all everyone will eventually become resistant to these therapies either in a short or a longer period of time. So there's something fundamental is happening in the prostate cancers that is enabling them to adapt and become resistant to therapy. And there's a lot of research really focuses in this end of the disease. So the very advanced cases, people are trying to find new drugs to try and uh find a way to to overcome these treatment resistant cells. Uh but I kind of consider this a bit like trying to plug the dam in the sense that just when you found a new target that seems to be quite effective, the cancer is able to evolve and grow and find another way to survive. Uh, and so rather than sort of trying to constantly plug all these leaks and every time there's a leak, it's more aggressive and more difficult to treat, my lab's taking a bit of a different view. So, we're keen to actually look back in time again here. So, we want to try and see whether there's a way to intervene and treat those patients with much higher risk of aggressive treatment resistant disease before it gets to this point. So before the horse has bolted.

[00:15:00]Now this is pretty challenging. This is not the kind of research that's easy to do because prostate cancer is a very long-term uh disease. And we also take a difference in the type of molecules that we're interested in. So again, a lot of research in this area focuses on genes, proteins, RNA. You've probably heard a lot of those terms before. But in our case, we're looking at a different molecule and those are lipids. Now, many of you would think of lipids and you think of when you go to the doctor and you get your lipid profile done in your blood. These are exactly the same types of lipids. But what you might not know is that lipids are also the building blocks of our cells. So all of the walls of our cells, all of these structures within a cell are made of lipids. So they're our building blocks, our Lego blocks that make our cells. They are also a major source of fuel for the cells. So they're incredibly important for cancer cells to be able to survive and thrive. So we reasoned that perhaps if we could look at the lipids in cancers, we could sort of see whether there might be some differences which indicate or predict the future behavior of that cancer because we know that lipids are so fundamental to the biology of a cell.

[00:16:15]So how do we do this? Well, we were pretty ambitious. We uh developed an international collaboration uh called lipids and prostate cancer where we got experts together that were focused on either lipid metabolism, endocrinology, medical uh oncology, uh prostate cancer biology, all different slightly different expertise coming together to try and profile lipids in clinical specimens. And we would never have been able to do this if it wasn't for a really unique resource that we have here in South Australia. And that's the Australian Prostate Cancer Bio Resource or APCB. So this is a bio bank that we've been running now for about 15 years. And I'm very privileged to direct that along with Samira, our tissue collection coordinator, and our amazing research nurses, Cass and Maud. So these wonderful women uh consent patients.

[00:17:06]They uh collect the specimens from patients undergoing surgery for prostate cancer here in Adelaide and they store the specimens but they also most importantly do waves of clinical follow-up. So we know what happened to these patients with their permission down the track. So with this we're able to design a really unique study where we could collect um the patients from our our bio bank look at their surgical specimens but divide them into these two categories. So the first being patients who never went on to experience a relapse and happily that's the vast majority of patients but there is a small subset who do go on to experience metastasis andor death from prostate cancer. So we wanted to compare these two groups and look at the patterns of lipids in their tissue.

[00:17:58]Now this little diagram here you might be wondering why are you taking all those microscope slides as well as the tubes of tissue. Well, the reason for that is that another innovation that we've really been building here has been to do spatial analysis of these prostate tumors. And there's a very good reason for that. So the cartoon I showed you earlier was very much represented cancer as a ball of identical tumor cells.

[00:18:23]That's obviously not the case. So a tumor essentially consists of a range of different types of tumor cell. It can also often have benign or non-malignant cells in it. It also has this connective tissue that holds the whole thing together. It has immune cells that infiltrate the tumor as well as the blood and lymph vessels that really provide it with the nutrients that it needs to survive. So all of this is a critical part of a tumor. But again, it's a cartoon. What does a real tumor look like? Essentially, it looks like this. Now, even to many people in my own team, this is not necessarily something that is intuitive when you look at it.

[00:19:02]This is essentially a cross-section of a prostate tumor that's then put onto a slide, stained, and then we can look at the features of that cancer. Now you or I would not know what those features are but we're very fortunate that uh we have a a great uh partnership through the hospital research foundation and we have a pathologist a research pathologist in Sajansy called Alex Dolly and she sort of acts as a bit like the cgrapher of this map of the tumor and she can sort of point out regions that are cancerous and ones that are benign and this is just a a nice way of of shading that to show you and this is a perfect example example of a tumor where actually most of this mass is benign. So this this yellow gold color, it's only these small pockets here that actually contain tumor tissue. And obviously if you're trying to understand what the lipids are doing in those tumors, it's going to be very hard if you uh consider it all together into one big uh one big mixture. Um, and if you think about it as a bit of a neighborhood, you know, the tumor cells over here might have grown up in a very different neighborhood than the tumor cells down here. So, they might respond very differently to therapy. So, we often use the analogy that it's a bit like if you don't consider the spatial context, you're really being like making a smoothie in the sense that you know what fruits are going in there, but you don't know what size they are, how fresh they were, what was the relative proportions. We really need to keep that hisytological structure. Now, this is probably the most scary slide that you'll see in the talk. So, the way that we measure lipids across a a tissue section is using mass spectrometry imaging. And this is a machine that fires a laser across the surface of our tissue. And that liberates all of the lipids and they go into this machine and that's then built up together into a map of how abundant different lipids are in different parts of the tissue. And of course we can then line that up with the hisystologology that pathologists like Alex can do for us. And one of the remarkable things that we notice this is the same tumor hopefully you remember it from a few slides ago that we saw before. When we look at the mass spectrometry results, we actually see that these different colors are areas that have very similar lipid composition. And what you can uh hopefully observe here is that these red areas have very different composition uh and line up perfectly with the tumor areas here um compared to the benign tissue. So tumor is already very different from benign prostate. But what happens when we take into account the long-term outcomes for these patients?

[00:21:49]Well, uh you might be surprised, I was somewhat surprised, but pleasantly so that there were evidence uh in these tissues that there are changes that we can see that line up with a poor outcome. So, for example, here's a couple of patients here. The top one was one that did not relapse. Here's one that did. Here are a group of lipids here where there's very little um in uh in the non- relapsing tumors, but suddenly we're getting a vastly increased abundance in the patients that go on to do poorly.

[00:22:22]And it didn't all go the same way because you're always a bit suspicious when things line up the same way. There were other types of lipids that were actually very abundant in the non-relapsing cured patients, but completely absent, lost when the p the that tumor went on to do poorly. So this was quite exciting to us because essentially what it was telling us is that uh tumors that go on to relapse are already metabolically distinct from those that are destined to do really well. And I just remind you that this might be decades even before the actual metastasis and progression occurs. So we were really uh you know gratified to see that these lipids are almost acting like a canary in the coal mine that actually something is going on in those cells.

[00:23:07]they're already primed to become treatment resistant. And so that leads us to a couple of different um avenues.

[00:23:15]So one thing we're doing is that we're partnering. It's a very unique cohort.

[00:23:18]So it's actually hard to find other similar cohorts worldwide, but we're partnering with the Walter Reed uh military hospital in the US who has a similar bioank and we're validating these changes to make sure it's not just us. It's not something unique to our situation. uh with the ultimate goal of creating a clinical test that will tell us hopefully in in greater certainty which um uh tumors are less likely to go on to become aggressive and those that perhaps need a greater intervention early. And to that end, I think one of the exciting potentials of the project that we're working on is to really use very advanced data analytical techniques to map uh the metabolic pathways that are actually causing these lipid changes because we think they might be more than just markers. They might actually be the drivers. And so if that's the case, we've already started uh identifying some new targets for therapy which we can work with people like Michael Roy to develop new drugs, but also um to try and repurpose therapies um because when you think about lipid modifying drugs, I'm sure many of you have heard those words before, they're very common drugs available for diabetes, for cardiovascular disease. And so some of those have the potential to be if we uh link them to poor outcome uh to be uh repurposed for cancer therapy. And so that that's a very exciting outcome. So our ultimate goal would be to have a a useful marker that will be a great achievement in itself, but if we could link it to a metabolic therapy that will be even greater. So that's our big challenge over the next few years. Uh and it's been a pleasure to present that example to you. So uh finally I'd just like to acknowledge my group at Sugency which is the prostate cancer research group and particularly Jacob Trung who was did a lot of the work that I'm presenting here today. Also our academic collaborators um our clinical collaborators without whom we could not do this research our funders but especially to the patients and their families. They consent almost unanimously to provide samples for the bio bank. they give their permission to monitor their medical records over time and as uh some of them as consumers participate in our research and actually give us their input to guide our our research strategies. So thank you very much for listening and I'll now pass over to Dr. uh to sorry professor I just want to make sure everybody hear me. All right.

[00:26:08]>> Okay, that's good. So, um, thanks Lisa for the passing the torch to me. Um, so I'd like to uh thank everybody for being here tonight uh both in the room and those listening um online and it's it's wonderful to see so many of you here tonight. So thanks for making the time.

[00:26:30]Um little bit about myself. Um my name is Lanwin. I lead the computational system oncology program at SGNC. Um I grew up in Vietnam. Um and then I trained in mathematics and computing in New Zealand uh all the way from bachelor degree to uh to PhD and then I spent a few years in Dublin in Ireland to uh horn in further my research skill set. um before setting up my own lab in Monas University in Melbourne. Um and then late 2024 um I moved to Adelaide to help build Sigency. Um so throughout my career um it's center on one important question which is can we use maths and computing uh to understand how cells make decision and can we exploit that understanding uh to our smart cancer.

[00:27:26]So again um uh the question of can we uh see cancers come back before it happens um is is everything that my team you know uh focus on. Um and as Lisa has explained it very nicely um cancer treatment can work beautifully at first right but um you know when you give the drug to to to patient uh a lot of the time the patients shrink um you know people feel hopeful um but then the scan shows something right um and you know maybe a few months a few years after cancer may come back um and so my lab is not only interested in how does the cancer look today. We're interested in how uh cancer might do you know tomorrow.

[00:28:16]Uh and uh this is at the heart of computational system oncology. Um and it's not just computing for the sake of computing. Um it is really about to stay one step ahead of cancer.

[00:28:34]So the the key insight is that cancer is not static. It's not just a um sitting target, right? Um when we uh cancer cells are living things and when we treat a drug, we create pressure. Now this pressure cause some cells to die which is what we want. Uh but some cells survive and the surviving cells can change.

[00:28:58]um they can switch on alternative survival pathway. They can find alternative fuel sources. Um they can become less dependent on the target that you actually target with the drug and all this is adaptation.

[00:29:16]Now the cancer isn't thinking of course right in the literate sense of thinking um but biologically it adapts um and adjust and this is one of the biggest reason why treatment fail.

[00:29:35]So um an analogy that I like to to use and I think is quite helpful not only for um myself but I think for other people is um imagine a spider web right so if you poke a strand what happens that strand doesn't just move on its own right because of the connectivity the whole web vibrate the whole web sift and the tensson distribute across all the threat And inside the cell, not only cancer cell, in every cell, protein work a little bit like this spider web. Uh you know, they are well connected. They are very complex. And a targeted drug is designed to hit one protein. It's like poking one specific part of this spider web.

[00:30:31]And the cell doesn't just respond at that one point, right? The the respond ripple through the network.

[00:30:39]And this is why we can't understand cancer just by studying a single molecule. We have to stand we have to study cancer as a as a interconnected network as a system as a whole. And that's why is the word system in my program which is system computational oncology.

[00:31:00]Now I think that's why the web analogy is useful but I like to show you what it like it look like a little bit more you know real um uh uh in the cell um so what you can see here is like a kind of a still abstracted diagram of the protein network within the cells but you can start appreciate the complexity um you see uh you know circuits that are responsible for uh proliferation which is how cell growth you you see circuit that responsible for for for migration differentiation and so on and all these pathway talk to each other they don't work in isolation and uh you don't uh you need to you don't need to read any of this right but um I just want the main point of this is for you to feel the complexity inside the inside the cancer cell So Lisa has talked about drug resistant that's the main theme of today's uh lecture but there's different mechanism of resistant right when people think about resistant people tend to think about new mutation but that's not the only mechanism some resistant happen rather quickly um through network rewiring through this this kind of spider web analogy when you poke a uh a note the whole the whole network adapts And we call that phenomena u adaptive resistance which is different to the genetic change caused by mutation.

[00:32:41]So think of it like this a cancer cell uh if when you treat the drug you shut the door right and cancer cell use an alternative door it's it use a side door to escape the drug target.

[00:32:59]So this is one the the central problem that my lab studies uh how do cancer cell rewire themselves uh under treatment and and importantly can we predict which side door the cancer cell might use next after treatment.

[00:33:14]Um but you might ask why can't we just know if we know the network diagram why can't we just figure it out right um can clever scientists just look at the diagram and uh and work out the answer now the answer is not always because this network is not acting linearly they are incredibly tangled together right um and pathway talk to each other, they feedback, they compensate. If you block one signal, maybe another signal might get even stronger.

[00:33:57]Uh the obvious treatment might not be the best treatment. The two drug that look logically might not work best together.

[00:34:06]Right? So, so these are the complexity.

[00:34:10]This is where intuition alone is not equipped to understand.

[00:34:16]And so what does this mean? This means that the cancer problem is bigger than any single person can solve in their head. And we need new tools that can handle the complexity for us.

[00:34:31]And this is what my team is building is new tool.

[00:34:36]So our approach is a cycle. Measure, model, predict and learn and test and learn. Um so there are there are four step this is a very simplified diagram but step number one is we measure how cancer cell behave in respond to treatment. What signals go up? What signal bounce back?

[00:34:57]What pathway turn on?

[00:35:00]And then we build computer models that describe the network in the cells.

[00:35:11]We build a model that can answer the question that experiment alone is very hard to to answer. Once we have a model, we can make new predictions. We can make prediction if we shut down this main door, what are the escape route that the cancer would take.

[00:35:30]And then we go back to the lab to test the most interesting and promising predictions. So we shortcut the the the experimental effort and this cycle happen over and over again.

[00:35:43]More data come in that have refined the model. So some prediction might be correct which is good but a lot of the time they might not be correct because uh we need more data or because our uh model is not yet uh perfect right and so the cycle come in more data generated but over time the model is refined our prediction get better.

[00:36:12]So if cancer escaped through side doors then the logical respond would be shut down a few doors at the same time and this is the rationale behind combination therapy. Probably many of you have heard about combination therapy which is probably one of the promising path in in terms of overcoming resistant that caused by by single drug treatment.

[00:36:42]So you shut down the main route with drug A and then you shut down the escape route with drug B. A simple idea, right?

[00:36:52]But in practice that that's not easy.

[00:36:55]Incred incredibly complex because there are many drugs available. There are many different pairs, different doses that you can combine.

[00:37:05]Every patient is network is slightly wired differently.

[00:37:12]And so it's not easy to choose which pair to give which patient at which doses and and we cannot test all these combination experimentally that would be very very expensive and very costly.

[00:37:30]So the scale problem is illustrated in this slide.

[00:37:36]We have too many drugs. So imagine if you have just let's say uh 20 drugs you might have already hundred of possible pair wise combination. If you look at three drug triplet triplet combination and you look at uh different doses combination different ordering the number explode exponentially and the number could be in in thousands and obviously when you have thousands of combination no lab can test one by one.

[00:38:06]uh but you know it's take forever and also it's very very expensive.

[00:38:10]So we need predictions to guide the experiment and this is why computation which underpin this prediction really matters. It is not because computers are more clever than our scientists, human scientists, but it because we can explore vast number of combination, vast number of possibility that no lab could ever test one by one.

[00:38:50]So now uh using that modeling approach in our lab, one of the uh most surprising lessons that um uh from our modeling work is that the question is not just about which drug to combine but it also um an equally important question can be when to combine the drug because timing can matter here. So in this guy diagram here you see scenario one where both drug A and B are combined together at the same time. Um you know it helps the tumor string. But if you treat the drug uh A first which is the scenario three and B after sometime you might better suppress cancer string and you might even help with toxicity.

[00:39:41]The other scenario when drug B first and I might not be as good as A and then B.

[00:39:50]And so same drug, same doses but different order could make dramatic different in terms of outcome. And so our modeling work in our lab actually show that there's that that that that could uh could be the real case under certain circumstances and we can predict when this might happen.

[00:40:11]And that would open up possibility to combine drug that concurrent therapy historical concurrent therapy might not work but if you combine them in a smarter way it might actually now work and that's another way of repurposing drug that thought to be not working in the past but you but then you ask why giving it in sequence might be better than concurrent.

[00:40:37]Now we know that and we actually have evidence to show that drug one push can push the network into this kind of vulnerable state a temporary vulnerable state and that's open up a window right a therapeutic window and if you treat drug B at that time when the cancer is weak then actually it's better treat B too soon might not work but too late. It might be already past that window, right? And so this timing is important and but but intuitively predict that timing is is really really challenging because not only because the network is complex that I show you but also because uh the timing could be very different between patient to patient. And so with model we can and with the right data coming from the patient we can try to get a sense of what that timing is between patient A and patient B.

[00:41:43]And so model can help identify the best window.

[00:41:48]So this means that one day choose not only which drug to give but als but also when to give it become important. A and as I said the timing here could be the different between resistant and maybe remission.

[00:42:06]So this kind of sequential therapy is one of the very exciting direction that we are uh working on in our lab.

[00:42:16]Another idea that uh we are also pursuing in in our lab that that follows this kind of network thinking um is called multilos therapy. So we know that traditional cancer treatment normally try to target cancer at uh uh as hard as possible. It's like a hammer approach, right? But when you combine multiple drug at high doses, the toxicity, the side effect can be quite severe.

[00:42:46]Um but from a network perspective, there could be another strategy, right?

[00:42:54]If you think that cancer depends on multiple route to survive, maybe we don't need to shut down completely a main one main route maybe maybe a possible alternative strategy is to shut down multiple route and apply gentle pressure to multiple route.

[00:43:14]So that can control the cancer but at the same time reducing harm effect harmful effect to the normal cell.

[00:43:25]So the aim here is not to smash the system. It's to control the network um in a way that can survive it hard to escape and reducing toxicity. Now that's the second part there of the illustration where you kind of try to control the network at multiple nodes.

[00:43:46]Now this is still early stage research um but it's a direction that that I think come directly from the the system thinking thinking from a network perspective which is also something that we are uh very very excited about.

[00:44:04]So over the the the past 10 minutes or so I've been talking about models that build from biology from understanding of cause and effect right when you when you block one pathway what what happened to another pathway but another powerful tool is coming into the picture artificial intelligence and probably many of you have heard about that particularly over the last few years AI is transforming many fields and biomedical research, cancer research is no exception.

[00:44:36]AI can analyze medical images, a lot of a lot a vast amount of patient data set. It can identify molecular signature that may predict how patient respond to a treatment.

[00:44:54]Um, but there's a important nuances.

[00:44:58]AI can be very good at spotting patterns in the data but it does not or it's not yet very good at explaining why those pattern exist and to me I think the why is actually very important in in in cancer research particularly when we study a disease that adapts to treatment.

[00:45:26]So we have here two modalities of kind of approach. One is AI which is like a tool that have incredible eyesight that that can spot patterns in the data that we human may miss. And we have this kind of mechanistic modeling that our lab is building that is based on cause and effects.

[00:45:54]But neither is enough on its own. And we think that the best is to actually combine this approach and that is one important direction that we are working on in the lab is to combine AI and this kind of mechanistic modeling to make better prediction and that lead to better and smarter way of doing experiment as well.

[00:46:20]So some people might think that AI or models might replace human but my view is that a computer does not replace a doctor but instead it give doctor a better map of what cancer might do next.

[00:46:39]And the the final analogy I would use today and I think maybe the one of the most useful one tonight is that of the weather forecasting. Okay. So, I'm sure most of most of us used it before we left the house this morning or maybe even last night. When we forecast the weather, we don't just look out the window. We collect data from satellites, from weather station, right?

[00:47:08]Thousand and millions of data points.

[00:47:11]What many of you might not know is that we feed those data into mathematical models. the kind of model just like the model that we are building and then we make prediction about the likelihood of raining and and sunny being sunny tomorrow right the more data comes in the model become better now I'm not saying that weather forecasting is perfect right but it is much much better than just guessing that is something that I think everybody have to accept So what we are moving towards in cancer research is a bit like that is taking the same direction in weather forecasting but for cancer forecasting.

[00:47:57]And so our hope is to be able to one day forecasting how cancer evolve when and how it might become resistant just like we do with the weather.

[00:48:11]But that's a long journey. Uh my feeling is that um where we are is about like where we are with for weather forecasting in 1970s right but that doesn't mean that we not we're not trying so why is incredibly uniquely positioned for doing the kind of forecasting that I'm talking about because we have experimental program the program like Lisa leads and we have computational program that I head together under one roof. And that togetherness under one roof is not only rare internationally, let alone in Australia, but is powerful because we can talk to each other on a daily basis.

[00:48:56]The data that Lisa produced can fit into my model and the prediction that we make can be validated by Lisa's team. So that cycle is important and not everywhere in Australia we have that set up.

[00:49:11]So this is my last slide semi- last slide which is about a larger vision of cancer forecasting.

[00:49:18]Uh we know that cancer come back because it adapts. Uh and we know that network can help can help us explain that kind of adaptation and we also know that models the mechanistic model that we build and AI may have forecast cancer's next move. So the goal uh is our goal in our program is to shift from reacting to relapse to anticipating resistance.

[00:49:49]Um and we made some good strike in that in that uh journey but there's a lot of work to do. Um, so I'd like to thank the team.

[00:50:01]This is the our program which is the picture taken about almost a year ago.

[00:50:06]So it's not the most upto- date. The team is about double of its size by now.

[00:50:09]I apologize to the team for not taking the most uh um uh updated picture. Uh but we can't do without the team. Uh that includes all the re research fellows uh um and and the students who are working incredibly hard every day.

[00:50:24]Um and uh yes, I would like to thank you for listening and I welcome um and looking forward to your questions. Thank you.

[00:50:39]So ladies and gentlemen, um please join me in thanking Lisa and Lan for such enlightening lectures.

[00:50:52]I think it really speaks to not only the depth but the diversity of um research going on at the institute. So before we proceed to the Q&A, I'd just like to um or just allow me to share a few words of thanks to the um community. So, Sajensi employs as you can see from Land and Lisa's talks uh a truly collaborative approach to cancer research and care and we bring together insights from clinicians, clinical and biomedical researchers, state and federal decision makers as well as the broader community.

[00:51:23]So if you have an interest in or lived experience with cancer um individuals, family members and carers uh can all join uh us as our community research advisory a partners advisory group. So to keep updated on our latest research and also to partner with us to drive meaningful research and outcomes.

[00:51:46]So, we're also very grateful to our philanthropic partners, donors, and supporters who help to fund our research. Um, it's certainly the case that we truly can't do this without you.

[00:51:57]Uh, cancer research absolutely takes a community, and if you haven't already, we invite you to join us um on this mission. So, I now open the floor for questions. Uh, we have roving microphones uh around in the theater. So for our audience on live stream, uh you're most welcome to ask your questions also via the chat box. And so at this point, may I please invite um Professor Christopher Sweeney um Sensy Director um to join on stage for Q&A alongside Lisa and Lan. Thank you.

[00:52:41]So, we have one question down here in the in the front that caught my eye. If the microphone is just on the way to you.

[00:52:58]>> Yeah, I've got actually two questions, one for each person.

[00:53:09]Um with regard to the modeling um is there imological modeling as well like what's happening to the immune system? Is that incorporated in the overall model?

[00:53:22]>> Is there can you can you um can you get it closer a little bit?

[00:53:26]>> Is there modeling of the immune system as part of the bigger plan for modeling?

[00:53:32]>> Yes. So uh so there are different type of different scope of the modeling that we can do. Um so we can a lot of our work is modeling what happen inside the cancer cell. So this signaling network that you see that's inside the cancer cell. Um but we also now doing a different type of modeling called agentbased modeling that treat each cell or cell type as an agent. And you can model the interaction between these different agents using mathematical laws right and rules that you learn from biology. So some of these agent could be just tumor cancer cell.

[00:54:10]Some of these agent could be immune cells in the stroma in the tumor micro environment. And actually we have a PhD student just starting a project in a lab that is looking at that kind of cancer and immune kind of interaction and use asben based model to understand that interaction. So, so the answer is is is yes.

[00:54:33]>> To Lisa, I'm just wondering whether the test you're working on, would that ever be able to be a blood test?

[00:54:41]>> That's Yeah, that's something we're definitely working towards. So, previously we've have been able to develop some um blood-based uh lipid markers for um for patient prognosis in prostate cancer. Specifically from these uh tissue ones though where part of that validation process that I talked about with the independent um sets of samples we will be looking in blood and tissue because yes obviously blood is going to be a lot more straightforward to analyze than than a tissue.

[00:55:18]>> Absolutely. Yes. Absolutely. Okay. So if you have a question uh just raise your hand up high and and we'll bring the mic over. But I think the next question for was from this gentleman down the front.

[00:55:29]>> Oh, thank you. It is a great presentations from both professor Lan and Professor Lisa. I have two questions. One each like I'm wondering I know your research is only after the person is diagnosed with the cancer. Is there any prediction that model you are talking can predict who will get reach get cancer or before the cancer diagnosed can we predict one second with Lisa I'm asking you talked about lipids or fats something that does it has uh any relationship with the diet throughout their life what fat their intake is there that leads to the that composition cancer fat composition or the is it the cause of cancer due to diet or fat that kind of things I I was I'm interested but maybe not your uh in this research topic it is thank you I'll guess for land first so I think the the question was regarding um models for predicting cancer predisposition in general >> right so I think that the question is you you know what I'm talking about is predicting whether cancer may become resistant right to drug but what your question about is whether we can actually predict cancer occurring even before it occur um so so uh we are not kind of so that's more like diagnostic type of of of of you know uh prediction um uh we are not working uh in that area uh uh with with a lot of focus but I know that other group in the world is doing that as well. Um so in principle if you have the right data uh good quality and and good amount of data to train this model right and particularly here I'm talking more about AI model um then they capable of detecting the pattern in the data of the patient before cancer and spotting what pattern might lead to cancer eventually.

[00:57:42]Now having said that I want to make it clear that it doesn't mean that the model will predict precisely which patient might get cancer or not but it's a bit more like think of it like a weather forecasting. weather forecast is not perfect, right? Um uh you know we we we can still get wrong when the cyclone come come in and and and then what is a part of the cyclone but uh it allow us to have a good probability of what what may happen and I think for for the mathematical predictive or even AI model that we're building it's a bit like that. So, so um it it give doctor like an extra uh decision kind of making ability that is better than guessing um that that that uh complement their expertise.

[00:58:34]>> And um the question for me was really around would dietary factors such as fat intake and the type of of fats potentially influence the lipids in the in the tumors. And the answer is yes.

[00:58:45]And so that's something that it's part of the reason actually that we're kind of quite interested in lipids and why we think they might make quite good biomarkers because they sort of they reflect the genetics of the cell but they also reflect the environment as of the cell as well which is our bodies. Um and so that's where um yes dietary intake will um inevitably affect not only the lipids in your blood but potentially the lipids in many of your your organs and and and tumors as well.

[00:59:14]So for example, we are involved at the moment in a few clinical trials where it's now becoming a bit of a a theme in cancer to do precision nutrition and so interventions in clinical trials that feature specific supplementation of certain types of fats and uh to see whether that has any impact on you know whether people respond better or worse to a given therapy. So I think we're going to see a lot more of this in the future because there's a huge amount of interest in it and and its potential to really um regulate how well someone responds to a particular drug.

[00:59:53]>> Okay. So we have one online question.

[00:59:56]I'll just hand over to Katherine >> advances in technology. I can pop down and read the question if that's helpful.

[01:00:14]>> Uh it's a question from Liz um online and does diet have an effect on the recurrence of cancer?

[01:00:22]>> Sorry.

[01:00:22]>> Any other questions?

[01:00:23]>> Sorry. Say that one more time.

[01:00:25]Uh does diet have an effect on the recurrence of cancer?

[01:00:30]>> Recurrent. Yes.

[01:00:31]>> Oh, sorry. Yes. So, um again, the um there's been a number of studies that um that do implicate that. Uh having said that, you know, it's the same problem with a lot of um you know, population-based studies is that the direct intervention type trials have not yet necessarily been done that would answer that question. And in the case of prostate cancer, it's really difficult because um generally prostate cancer is a very slow growing cancer in the sense that to get outcomes from those sorts of studies is going to take many many many many years and so a follow-up and so that's where some of those studies get really really challenging. So a lot of the studies that we have are epidemiological studies. So it's looking back in time, you know, so sort of, you know, looking at people's um dietary diaries or sometimes even blood tests that are taken along the way. But again, they're largely associationbased studies just purely for logistics, I think. Um but, you know, I'm excited by the the prospect of some of these new trials where there are short-term interventions with a specific dietary um intervention.

[01:01:39]And you know, I think that will finally give us some more direct clues and more direct evidence as to whether this is a you know, direct association or whether it's actually something indirect going on. I'll just add there was recently a clinical trial in colon cancer with exercise which modulates your metabolism and improves your health system where patients who had chemotherapy for cancer that had spread to the lymph nodes and fully rected with exercise versus chemotherapy without exercise and they had less recurrences and actually had an improved overall survival. So it actually does speak to a whole host between the metabolism, the immune system and exercise impacts both of those. So while we may not have a particular diet, microplastics or whatnot, but we do definitely have strong implication around the exercise and the nutrition associated with that.

[01:02:26]>> Agreed. And I think there's been very good data coming out in breast and prostate cancer on that exercise front as well in terms of you know preventing a lot of the side effects of hormonal therapies and um but also showing improvements in actual outcomes too. So I think that's a space that's going to be very exciting to watch. Sorry. Okay.

[01:02:43]So, we might have one last question from this young man in the middle and then um >> there'll be there'll be some opportunity also to speak to the speakers after after the event. So, we'll just have one last question and then and then we might draw it to a conclusion.

[01:02:58]>> Well, thank you um for in increasing our education on a very important topic. Um, my naive simplistic use of AI so far has illustrated quite amply that it makes mistakes.

[01:03:14]My question would be, how do you know that your AI assistants are not making mistakes? And I think perhaps you might have answered this, Professor Noyen, just a little earlier in part by saying that you have a a closed loop system, a trial and error system, which is okay if it's a tight system.

[01:03:34]Um, weather forecasting is pretty tight.

[01:03:37]I mean, you know that something's going to happen within days of when you predict it, but in the case of cancer, as you've described, it's years. So, how do you know that AI is not leading you up the garden path?

[01:03:54]Uh so I I guess your just to summarize the question would be we know that AI can make mistake they can hallucinate.

[01:04:03]Uh if you uh have some experience with chart GPT which is something that we heard about a lot nowadays they do hallucinate. Uh so the question is would they hallucinate in the context of weather forecast uh cancer forecasting?

[01:04:17]That that that would be your question.

[01:04:21]I would say that uh uh our model make mistake all the time when we when we develop and testing them right uh because when we compare the model prediction one way to test whether the model is performing is to compare the experiment against the prediction of the model um and that's a cycle that I was talking about when we make prediction um that are interesting and I think worth uh doing in the lab we go back to the lab and test those predictions. So we treat cancer cell with drug to see whether it actually behave like the model predict and that is the best test in terms of testing the the model. So sometimes as much as we hope the model does predict correctly but more importantly a lot of the time it doesn't but that's actually not a weakness of the approach is actually a strength in a way because when you see discrepancy between the model prediction and the experiment you know something is wrong. We know that our model is not accurate and so we go back to our model and identify what is missing in the model because this model are basically a computer replica of our understanding.

[01:05:40]So we build our understanding into the model. If the model is wrong, that means our understanding is also wrong and that is the actually an incredible you know valuable opportunity to go back and update our understanding by finding out where the model is wrong. What is it missing? And so a lot of the time when we go back and find out oh maybe it's missing a feedback loop, maybe this pathway that we supposed to to have in the model is not there in the model. We put that in right and the model can perform better. Sometime that pathway is not even found in the literature. We know that something is missing because when we add that it explain the data better but it's not yet reported and that is actually an opportunity for us to find something very novel and so um the answer is that this model making mistake but we are doing a lot of rigorous experimental work to validate our model right and I think that's the best test just like chip is making mistake but because more people using it right more more data generated those data fed back to the company that developing CH GPT and if you compare GPT nowadays and say three years ago the the hallucination is dramatically reduced and so again back to the point that the model are not perfect but they are becoming better over time as we have richer data we have better quality data I have more data >> okay so I'm advised that we can sneak in one more question.

[01:07:18]>> Oh, sorry.

[01:07:20]>> Can you hear me?

[01:07:21]>> Yep.

[01:07:21]>> My my question is for Dr. Butler. The Gleason score is used today to uh estimate progression of cancer in the future based on microscopic examination by a pathologist. Uh any correlation in your studies with the lipid content of membranes and the gleon score?

[01:07:37]>> Yeah, that's and I'll follow up.

[01:07:39]>> Yeah. Yep. And uh is it possible that your technique may improve or or replace the Gleason score as the best predictor of future progression?

[01:07:48]>> Yep, that's a great question. So just for the audience, um the question was um how does this uh these lipids and the signatures that we're developing, how do they relate to the existing Gleason score, which is the main pathological scoring system for prostate cancer and maybe will the lipids um build on or replace that kind of score. So the answer is yes. There does seem to be a correlation between the abundance of some of these lipids and the Gleason score. I think what we've tried to really focus on is the gray area of prostate cancer. So when um the Gleason score is very low or it's very high, the um the treatment path is a little clearer in the sense of when it's very low, it's tends to be more conservative management now. Um, and if it's very high, we know that that's actually going to have quite a high risk of recurrence.

[01:08:36]But there's a huge gray zone in the in the middle. So, Gleason 7 cancers for those who are familiar with the um the grading system where, you know, it's really very very difficult to tell which ones are going to do well and which ones are going to do poorly. And so, what we hope is that I doubt it will replace it, but it we hope that it will add predictive value. And in fact the biostatistitians that have been working with our lipid data have shown that when you control for existing clinical factors including gle gleon grade these do add extra information. So extra any bit of extra information I think is is useful to have and so that's the goal is it would be sort of a multimodal type of test or signature in the long run.

[01:09:20]>> Okay. So it's great to have so many questions. I think we've got one more question we can actually fit in. So this gentleman, did you Yep. We've got one more in the front.

[01:09:29]>> Question about prostate cancer.

[01:09:33]>> Oh, the microphone's just on the way.

[01:09:34]Just a moment.

[01:09:36]>> What with prostate what's happening in the 10% prostate cancer victims that don't survive? Why is that? The other the other point is um people with higher cholesterol, are they more likely to get prostate cancer? And the third one, I wonder if this is true. Um they reckon Asian men are less likely get prostate cancer than African men. If that's true, why is that why is that the case?

[01:10:00]>> So, I reckon that's a sweening question.

[01:10:02]>> This to Lisa or or Chris, but Chris is the best qualified.

[01:10:07]>> Uh the what we do know is that the the men who die of prostate cancer about twothirds of them relapse from local disease because that's so many patients present with localized disease. Having said that, that number is starting to come down. So five years ago the prostate cancer specific survival was closer to 85% chance 15% died after local therapy in the high risk. We recently completed a study called Enzad where we've decreased that number down to 3% dying at 8 years of prostate cancer. So I think that number is going to go down further and that's because we've identified it earlier and got the more effective therapies which are more effective when the cancer is less resistant. the other group of patients who die of prostate cancer, those who present with large disease that just grew despite PSA screening. And there's something abnormal about why those patients got cancer. Nothing that they did or had ingested. It's just something unique about that rare. And that's only 5% of the prostate cancer patients. So our job is to identify who gets prostate cancer with despite PSA screening. And that's where we need to break down that down.

[01:11:23]To your point about Asian versus African versus Caucasian, there are some clues.

[01:11:28]So once you take out some of the features related to socioeconomic features of uh people living in poor areas, there does seem to be a different biology between prostate uh Asian and African and Caucasian.

[01:11:43]And that's probably a genetic basis.

[01:11:46]There's something about the genetics of men in certain parts of Africa. West Africa has a different biology to men from prostate cancers from East Africa.

[01:11:54]So it comes down to genes more than anything. And comes to the previous question about uh risk factors. How can we identify a man's general genes and even women and their risk for are given cancer. So it's their given genes that they born with paired with their environment that's leading to it. and Africans, men of African descent, men of Asian descent, Caucasians have different gene backgrounds that may contribute to that.

[01:12:20]>> What about an association between high cholesterol?

[01:12:25]>> I'll I'll put that I'll not be good.

[01:12:29]>> I'll put that in another cholesterol.

[01:12:31]>> I'll answer that. We can decrease 50% of all cancers if we get weight under control, no smoking, and decrease alcohol. All the preventable causes. So that actually speaks to all of the things that we measure, cholesterol and sugars and weight. So the answer is if you can control all of those, you'll decrease your chance of cancer by 50% for all cancers.

[01:12:54]>> Okay. So >> a public service announcement.

[01:12:58]>> So I think um with that I'll just say thank you to everyone for your active participation in tonight's uh lecture.

[01:13:05]Thank you so much to Lisa, Lan, and Chris. Um our speakers will stay on for a few more minutes after this um the seminar uh at the front to answer any qu additional questions you may have. Um so it's great to see so many questions from the audience. Um thank you everyone for joining us tonight. Please uh enjoy the refreshments that our team have prepared for you in the foyer and feel free to chat with our Sensy um early career researchers and PhD students. Uh, so you'll be able to recognize them because of their big smiles and because they're wearing bright red buttons that say let's chat.

[01:13:39]So just Yeah. So just one last thanks. Um, thank you everyone. Uh, enjoy the rest of the evening and good night.

[01:13:59]You did it.

[01:14:03]>> Thank you.

[01:14:27]for sure.

[01:14:46]What's happening?

[01:14:49]There's something interesting informative at the moment organiz