 Thank you. My name is Sarah Kate. I'm a PhD student in the United States. This is my second year at CSVConf. I went to the version four in Portland in 2019. Whoa. Sorry about that. That's me with the Kamalama. I am a PhD student studying cell biology. So as a cell biologist, my talk is going to be a little bit different than a lot of the other CSV talks. I'm going to show you data, but it's data I got from the lab, not from mining or coding. But I think there's a lot of overlap between the type of projects I do and the type of projects you do. And I'm hoping my talk will have some lessons that are common to all scientists working with all types of data. What I'm hoping you're going to get away from my talk, I'm going to cheat and tell you the moral of the story in the beginning. I hope you're going to learn that there's a lot you can do with a really small amount of data, as long as you're very careful about it. And when the data doesn't go well, it's OK to pivot, even when that feels really scary. And finally, that scientific stories are never over, and things can come back in really unexpected ways. So the title of my talk is about child to dementia. I know that's a really provocative, scary-sounding phrase, but it is a real disease. The actual name of the disease is Neiman Pick Disease Type C, or just NPC as it's abbreviated. It mostly affects children and young adults, but you can get presentation at any age. In very young children, this looks like failure to meet developmental milestones. As you get older, this can look like behavioral problems. And then it does evolve into a very dementia-like phenotype where you forget things. There's cognitive decline. Something I really want to emphasize about NPC is just how rare it is. In the entire United States, there's about 960 patients. If we extrapolate based on population data, we'd expect about 130 patients in Argentina. This is a very, very small number. And not all of these patients even are diagnosed or coming into contact with clinicians. So the actual number of patients and, I guess, data points that we have for this disease is really, really small. And this can make answering even very simple questions very challenging. So a question you might have is, what is the average life expectancy of an NPC patient? And I apologize because I think my collars aren't totally coming through. This seems like a very simple question. And we do have an answer. We think it's about 14 years old, that 50% of patients living that long. But to get this answer, the researchers behind us had to do a lot of work. They crowd sourced information from disease support group websites and memorial walls to find out how long those patients lived. They had to do this manually. There's a ton of work. And they still only found about 338 different people. And so even just getting this very simple answer was very, very challenging. Another question you might have is, what treatment options exist for NPC patients? And unfortunately, there aren't a lot of options. There's one drug that's been improved in Europe. And then there's this other drug called cyclodextrin. You don't have to remember that. What it's showing you here is that this is an untreated NPC patient. And this is their neurological severity score. So this is the severity of their dementia. You can see it's pretty high. And then in the treated group here, treated with cyclodextrin, you do see that severity come down pretty significantly. This was a super exciting finding. They found this drug completely by accident. They found it in 2017. But it has really struggled to get approval for treatment. And it's partially because there's just so few patients for a clinical trial. And the disease progresses so slowly that it can be difficult to get statistically significant data. So what can we do? I think at this point, I've painted a pretty bleak picture. We have childhood dementia. It's a very short life expectancy, no treatment options, a little depressing for a Thursday morning. But there are options. Something that NPC patients need right now is a very good biomarker. One of the hugest problems with the disease is that it's very difficult to tell is a patient getting better? Is it getting worse? A very common biomarker you might be familiar with is something like a fever. If you're feeling sick, your fever goes up. You know you're getting worse. If your fever comes down, you know you're getting better. We need something like a fever that is very specific to NPC. Now this is my one cell biology a little bit technical slide, but this is what I work on. They're called extracellular vesicles. They're huge news in the cell biology world if you didn't know. And they're very promising biomarkers. This is a cell number one. And it's secreting these little purple dots, which are EVs, to cell number two. You can kind of think of extracellular vesicles or EVs as texts going between different cells, sending little messages back and forth. These EVs are full of different types of proteins and molecules, and the proteins and molecules that are inside of an EV reflect the cell it came from. So for example, if the cell had cancer, this EV would contain cancer like molecules. We could use that to help diagnose cancer. My question is, if a cell has NPC, can we use EVs to help track NPC? The way I started looking at this is with spinal fluid because remember, NPC is a neurological disorder. So your brain is in your skull, everybody knows that, but it's not just sitting there, it's actually floating in a lot of fluid. That fluid bays the brain and then it goes all the way down your spine and we can take it out of your spine and use it to get information about what's going on in your brain. In this spinal fluid, there's lots of different molecules. Some of them are gonna be EVs, which are these little purple dots. That's what I want. And so what I go in and do is separate these EVs from all the other stuff that I don't want. Once I've done that successfully, I can look at them under a microscope, a super fancy microscope, because these are like 10 times smaller than cell. They kind of look like they've got a nice membrane, they've got some cargo inside of them. Once I know that I have EVs, I can start looking at what that cargo is and seeing if it's related to NPC. So I started with only 12 samples from the clinician that we know. So we had six samples from NPC and six samples from healthy patients. Again, that's a really small number, but it's actually really big in the NPC world. I was really lucky to get these samples. And in the beginning I was blinded to all the metadata about these samples. I didn't know anything about them. I just had some NPC tubes and some healthy tubes. So I started looking at their EVs and I'm not gonna get into the technicalities of how all this happened, but please ask if you're curious. And I immediately noticed this strong trend where in my NPC samples, as compared to my healthy control samples, there's just a lot more, lots more EVs. And this was true for a couple of different markers. I'm just showing you two here. I wasn't expecting to find this. It was a huge surprise and I was super excited. I was a first year PhD student and this was my first big result. So I immediately started thinking of hypotheses and theories and I wrote our collaborator and I said, hey, I got this amazing data. I wanna know what does it correlate with? Is this a severe disease relation? Is this age of onset relation? What correlates with this? So she sent me back the data. Now I have all this information. I realize my cohorts were not very well matched. Oh well. I started trying to run correlations and see does disease severity affect EVs? Does age of onset affect EVs? The only thing that was statistically significant, unfortunately for me, was age. What I learned was that the younger my patients were, the more EVs they had. And this was a massive problem because I don't have young controls. Healthy young kids do not give spinal fluid up. They just don't. And so essentially what this means is that I can't tell if this very cool increase in EVs is coming from the fact that they have NPC or just the fact that they're about 10 years old. There's no way to know. So at this point, I was pretty devastated. Again, first year PhD student, very sad about things. I didn't really know what to do. So I thought to myself, okay, you wanna be a cell biologist. What would a cell biologist do? Probably look at cells. So I completely pivoted my entire project. I started working with skin fibroblasts. So this just comes from a skin biopsy. They're pretty easy to get. They're pretty easy to grow. They make tons and tons of EVs. So a little bit easier than working with spondyloid even. And what I wanted to know is can I replicate this trend in my cell model? The first technique I tried actually worked. This replicated really well. But by now I was a wise, bitter, cynical scientist. And so I didn't really believe this. I thought, okay, I'm gonna try another way. So I tried a different technique a little bit harder and a different way to count EVs. I'm still saw the same trend, but I'm still not buying it because I'm still really cynical. So then I came up with the hardest possible technique that you can do. Also the most accurate. That's why there's no stats on this because I've only done it twice because it's a very difficult experiment. But again, I see the exact same trend. And now I'm ecstatic. I have the exact same pattern as CSF. Everything is looking really good. Caveat being that this took me two years to go from this to this because switching models and ideas just takes a long time. And you might be wondering, like in the beginning you told us all about how this is the brain and we need a biomarker, but now you're in cells. Like is this even useful anymore? I had the same question. So I looked at the molecules that are inside my EVs, this part. And what I actually found is that there's a lot of overlap between what's in my cell EVs and what's in my spinal fluid EVs. So it's totally possible that we could use the data that I get from cells to help inform our next steps for spinal fluid. So that is the entirety of my talk. Just going back to what I mentioned in the beginning, this project started with a really small amount of data, only six patient samples, but I still managed to get a lot out of that. When my CSF data turned out to be kind of questionable, it was very challenging and difficult for me to totally pivot my ideas, but I managed to do that and actually worked out pretty well. And then this is my last kind of nice little cheesy note. Even though I sort of gave up on my increases in EVs story, it actually did come back to me in the end and was true in my second model system. So sometimes when you think you might have to give up on a scientific story, you might just have to persevere a little bit longer, like maybe two years. Also, if anyone in the audience is about 20 years old and wants to donate spinal fluid, I am still looking for that. So, thank you. Thank you so much, Sarah Kate. Has anyone got any questions? Straight in. Thank you. Great talk and this is like real science. It's really, really exciting. I have a question, but at the beginning, you talked about crowdsourcing data about patients. I'm curious to know, did you do that with consent or what was the ethical approach to that? Sure, so I was not part of that study. That was a study led by Dr. B. and Coney. And I think they were mostly pulling data from publicly available like memorial walls on patients that had already been deceased for quite some time. So that data was a little bit protected. What is really interesting now, pretty recently there's this effort to make like a centralized reporting system for all NPC patients. They've launched it in Europe, I think a year or two ago and they're trying to make it international with fully consented living patients. And so that would be kind of an ongoing data repository. So we wouldn't have to do things like scrape memorial wall data, but that's kind of just getting off the ground. Thank you so much for this talk. And as a non-scientist, thank you for explaining biomarkers and EVs and like the best, cutest way ever. That was fabulous. I'm curious if you have interacted with or collaborated at all with other rare disease data consortiums. Yeah, it's not something that I have done yet, but it's definitely something that I want to do. I spend a lot of time on the Nord website that like works with a lot of rare diseases. Other rare disease groups can be really helpful in terms of finding resources, but the diseases themselves can be so different and specific that scientific advice can be difficult to get. But one thing I will say is that as like a baby scientist, coming into rare disease field has been so welcoming. Everyone is just so excited to support you and help you because they want so badly for you to work on this disease. And so just being in that community of rare disease researchers has been really rewarding. Thanks for your talk. That was really, really interesting. So what's your order, your next steps going forward? Great question. I like so many ideas. My big thought is why are there more EVs from this disease? And I think there's two possibilities. Either more EVs get made or less EVs get taken up. And so they're more circulating. So I'm basically testing those two different hypotheses in a variety of extremely slow ways that might take more than two years. But we'll get there, yeah. Any other questions? No. Can I just say thank you as well? It's been ages since I've seen science and it's lovely. That's right. Great. Thank you so much for a wonderful talk.