Think Verizon is expensive? Think again. Anyone can bring their AT&T or T-Mobile bill to a Verizon store today and will give you a better deal. So bring us your bill. Walk in, run in, Pogo stick in, teleport if you can. Ride on the back of a roller-blading yak or fly in on the wings of a majestic Falcon. Any way you can, bring your AT&T or T-Mobile bill to a Verizon store today and we'll give you a better deal on the best network. Based on Rootmetrics' best overall mobile network performance, US 2nd Half 2025, All Rights Reserved, must provide a very recent postpaid consumer mobile bill in the name of the person who gave me the deal, additional terms, conditions, and restrictions apply. Close your eyes. Focus. Listen to work getting done with Monday.com. Relax as AI does the manual work while your teams are aligned on a single source of truth. Feel the sensation of an AI work platform, so flexible and intuitive, it feels like it was built just for you. Notice you're limitless. Now open your eyes, go to Monday.com. Start for free and finally, breathe. Nature and experiment. I really didn't know you. Why is it so bright so far? Like, it sounds so simple. They had no idea. But now the data is clear. I find this not only refreshing, but at some level astounding. Nature. Nature. Welcome back to the Nature Podcast. This week, the effectiveness of off-label cancer therapies and an ancient genome study, capturing human evolution in action. I'm Benjamin Thompson. Nature. First up on the show this week, we've got a story about a nature paper that's pulled together and published 10 years of data from a massive Dutch cancer therapy trial. Specifically, this trial has been looking at the effects of off-label treatments, where a cancer therapy is used to attempt to treat a disease other than the one it was designed and approved for. And there are reasons why drugs might be used off-label. Perhaps a person's cancer didn't respond to any of the approved treatments, so options are limited. Or perhaps they have a rare type of cancer and there aren't many treatment options in the first place. There have been examples where results from this approach have been positive. An estimate suggests it's happening with some regularity, although numbers differ from country to country. The decision on which drugs to try is usually informed by similarities between the types of cancer involved. For example, doctors might select a drug designed to treat a cancer in a completely different part of the body, but which shares the same genetic profile as the one in the patient. Perhaps they share a similar mutation or some other sort of DNA mistake that makes the cells susceptible to the drug. However, as off-label treatment tends to be a last resort option and one often done in an ad hoc way, data on outcomes is patchy. This can result in oncologists not knowing about other successes and risk someone who is already very sick being treated with a drug that others have found to be ineffective and which might have serious side effects. Getting data on what works could lead to changes in prescribing habits, better outcomes for patients and changes to things like insurance reimbursements to pay for treatment. And providing this data is what this week's paper is looking to do. It reports on a decade-long prospective trial following the outcomes of people with advanced solid tumors being treated in hospital to cross the Netherlands. The trial looked to capture how effective drugs designed to treat a cancer with a particular genetic profile were when used off-label to treat a different type. I spoke to one of the members of the team, Emile Vust from the Netherlands Cancer Institute. Emile is a practicing oncologist and he laid out why getting data on off-label drugs was important to him. For me, the driver was that I had a young patient 34 years old who had metastatic disease and no other opportunities with a fairly rare cancer type. Obviously, the impact of somebody of that age with a one-year-old child is huge and then you start searching for opportunities to treat. And then you come across that are drugs that are on the shelf that may work in a particular setting because they have the similar DNA mistakes. And for me, there was a starting point to say, OK, let's get everything that we potentially have because if we bring those types of data together, then we can really help patients. And obviously, those patients are at the end stage of their life because of the disease. They have exhausted the treatment options. So they're very vulnerable and treating patients and still having a lot of side effects that's not good healthcare. As a doctor, I have an obligation to protect my patients, but in the meantime, also provide them with the best opportunities. And that's where this work comes in. And you and your colleague set up a really big trial, the Drug Rediscovery Protocol. Now, this trial began back in 2016. Tell me about it. This is a fairly novel trial because it's a crossing of lots of different cohorts with lots of different treatments. So what we try to do is make little cohorts starting with eight patients of a tumor type, a DNA mistake, and a treatment. And we did that to make sure that we would have the context of the tumor included in our analysis. And then we said, if we see in those patients of eight, if we see an effect so that the patient's tumor regressed or that the tumor stabilized for at least four months, we said, OK, that's a success. And then we expanded with more than 16 patients. You go to 24 patients for stage two. And if we see that at least five of those patients benefited, then we go to stage three where we have a larger set of patients to kind of confirm the earlier results. With that setup, it created a lot of interesting challenges because there are so many different tumor types, so many different genetic aberrations, so DNA mistakes. And we had at the peak 37 different targeted drugs that we could use. So we had hundreds of cohorts that we started and some of them were very rare cancers. So then we brought all that information together and we could see that some of these cohorts actually worked really well. Patients lived very long after being treated with those agents. But unfortunately, the majority has very little benefit at all. So you had all these patients, then well over a thousand, and you've tried these different drugs working in different combinations with the different genetic cancer types. You say there that there were some successes, but also some things that didn't work quite as well. It does seem that the results were quite sobering. It suggests that the effectiveness of off-label drug treatments was modest. Could you give me the overall numbers? As an oncologist treating those type of patients, you need to stay a little bit optimistic. So I'm a glass half full type of person. So let's start with the good news of the 6010 patients that we treated. Roughly one third of the patients had either tumors that were partially regressing or they stayed stable for more than four months. So that's one part. Then out of those patients that we treated, 7% had a very long response rate of years. And some patients actually lived on for years and years and you could argue that some may even have been cured. So if you're a glass half full type of person, then you say those were patients that were not treated otherwise with drugs because it's not on label. They didn't have the tumor type that would actually be used for treatment. So that is very positive. But it's also humbling to see that 93% of the patients did not benefit that much, but still underwent the treatment with the side effects. And that's the reason why I think there are two big lessons here. First, we have drugs on the shelf that can truly help patients. Even if it's 7% or 30% depending on how do you estimate the value. And secondly, we also need to be humble in the sense that we cannot just give patients these type of treatments without considering the side effects. So we have an obligation to really share that data. And for me, the fact that it's now published is really important because we have cohorts of N is one or two patients with rare cancers. You never can publish all these case reports. So we thought, OK, we create one big resource that now the scientific and clinical community can also have access to. So you saw a really good outcome in about 7% of the patients. In terms of the results, do you get any sense of why this might be? Because the setup is that Cancer X and Cancer Y have a similar genetic component. And on the face of it, if a drug targets that specific component, then it would work for both. But clearly that doesn't seem to be the case very often. Any ideas as to why that might be? I think that's the key question. And if you would ask me 10 years back, I would say, well, if we know all the DNA, we can identify the vulnerabilities and bring drugs together. And then magically, the cancer would kind of be no problem. And now we're like 10 years further down the road. And cancer is a different beast and very good in finding solutions to go underneath the pressure of a certain treatment. Having said that, we still made advances. It's important to realize that these targeted agents that we use, the ones that are used on label, obviously have a very clear and significant effect. So those are not in our study. We're taking the rest of those patients. I mean, you do say that there were some bright spots. Has this altered treatment regimens across the Netherlands or any policies surrounding health care? One of the important things is once you have data that are clinical trial grade quality, you can actually use them to convince the regulator or the reimbursement agencies to kind of provide reimbursement for those type of drugs. So we succeeded to get more drugs into the Dutch healthcare system. And our study has been a model for many other studies in the world. So there are multiple studies in Europe now that we provided them with the whole protocol. The only thing that we ask you is if you set up your study, make sure that data sharing is an important factor. Because then we can also bring those data together with our data and then show that we have enough evidence that a drug may or may not work. And especially patients with rare cancers have no opportunities because it's not very attractive for pharma to develop drugs in these rare cancers. And this is a large trial, but it has to be noted there are limitations. For example, the trial began in 2016, which might only be 10 years ago in terms of time. But in terms of things like the progress in cancer treatments and in genetic screening, that's an absolute age. Do you think if you were to start the trial again today that you would see different results? I don't think it would have changed that much because many of the drugs are still available, are still around, are still very important in their unlabeled applications. Obviously, there will be new drugs and that's good because that moved the field forward. But I think in general, if you look at how these drugs that we have used are still available, then I would say they are still considered to be standard of care. Now, I know there's a big cancer conference coming up this week in the US and people from all over the world, I'm sure, will be there. What do you think they'll make of your paper and what questions do you think they'll have? I'm quite sure that there will be people who say, well, this is very disappointing because it hasn't brought that much and that will only help people to think in a different way about strategies to really come up with better treatments because we definitely need those. And then you will have people that say, OK, but we also have helped a significant amount of patients with these drugs that were already on the shelf. And let's make a infrastructure that perhaps can bring that data together. And then finally, I think people with rare cancers will really be happy with our initiative because people with rare cancers hardly ever have large clinical trials with new drugs, new opportunities. So they're always lagging behind the patients with the more common cancers. So this is a study that everybody can have an opinion on. So we're not trying to convince anyone. We just wanted to make sure that all the data that we have generated is now available to the public. And we also show our struggles and successes. Emile Vust from the Netherlands Cancer Institute there. To read the team's paper, look out for a link in the show notes. Coming up, an ancient genome study looking at natural selection in human genes. Right now, though, it's time for the research highlights with Charmony Bundel. The ice hockey career of E. coli may be about to take off as researchers demonstrate that these bacteria can spin round microscopic pucks without ever touching them, using only the rotation of their cell bodies and tails. E. coli propel themselves by spinning their corkscrew shaped tails in one direction and forcing their bodies to twist the other way. In open water, these rotations have little effect on nearby objects. But a team harness this motion by sandwiching the bacteria between glass and a microscopic 3D printed disc. As a bacterium swims into a channel on the disc's underside, its body stares the fluid in a way that generates torque, helping to spin the entire puck. Previous bacterial motors required carefully designed gears that the microorganisms had to physically push against to move. But this alternative mechanism is contactless and works on a greater variety of shapes. The finding sheds light on overlooked forces that may occur in soils and biofilms and paves the way for microscopic conveyor belts and other bacteria powered machines. Spin yourself over to that research in Nature Physics. What are the risks of dying early in Australia? Middle-aged women who are physically active on a regular basis can slash their risk of dying early in half, according to a study of more than 11,000 women in Australia. The World Health Organization recommends that adults should do at least 150 minutes of moderate or 75 minutes of vigorous physical activity per week. But past studies often focused on measuring activity at a single time point and typically couldn't establish whether activity causes improved health or if the two are linked by other factors. To overcome this, a team used data collected every three years from middle-aged women in the Australian Longitudinal Study on Women's Health. The authors relied on an analysis method that allowed them to use the data to emulate a randomized clinical trial. They found that the women who met the WHO recommendations for physical activity throughout midlife had about half the relative risk of dying from any cause than those who consistently did not meet the recommendations. Jog over to PLOS Medicine to read more on that. This week, Nature has a paper looking at the evolution of the human genome. And here to tell us about it is senior reporter Ewan Callaway who's been covering the story for Nature. Ewan, how are you doing today? I'm all right. Thanks for having me. Not at all. Thank you for being here. So let's talk about what this new paper was trying to figure out then, Ewan. It was looking specifically at homo sapiens, us, and how our genome has changed over time. Yeah. So let me just give you a crash course in recent genomics and studies of natural selection. And this is looking in modern humans. You know, the first thing people did in the genomics era was kind of look for gene variants that had telltale signs of natural selection. They were really, really, really common in one population or another. They had some other telltale signals around them that I won't bore you with. And doing that didn't turn up very many hits for signs of natural selection. I think the poster child, Europeans at least, is this gene variant that allows you to basically drink milk as adults. And people have come up with kind of stories is that it provided some advantage to people after development of agriculture and farming. And so in different European populations, you see this blazing red signal of natural selection for this trait called lactase persistence. There have been a few others. There hasn't been a whole lot. And the ancient genomics era, when we started getting ancient genome data from thousands of people, people thought, ah ha, you know, we're going to see evolution more clearly. You can drill back in time and ask, in ancient time periods, did one variant become more common than the other or not? And unfortunately, we really haven't found a lot of signals of natural selection in ancient genomes, even increasing the sample size. So it led to this head scratching notion that maybe we're in this kind of stasis. Maybe our genes aren't evolving through natural selection so much. I should say I'm using the term natural selection here as a catch-all. This paper is talking about one very specific form of natural selection. It's called directional selection, either positive or negative selection. And that's when a gene variant is so helpful that it surges in frequency. And if time carried on, everyone would have it. Or, you know, not having it is a benefit. So the gene variant disappears. So that's exactly what we're looking for, these examples of directional selection. And so if that's the setup, then, you, and what does this paper do specifically then to try and look at these different things? The study is laser focused on Western Eurasia, which is the region that includes Europe and the Middle East. And part of the reason is because that's where the vast majority of ancient DNA samples come from. So, I mean, number one, it adds data. Like it adds another 10,000 or so new ancient genomes doubling the size of the ancient genomics literature, you know, so it adds numbers. This is probably by an order of magnitude the largest human genome study ever. And it's asking this question over the last 10,000 years, were there gene variants that came under natural selection in humans from Europe and the Middle East? The other thing that this paper does, and this is really where the bone of contention is going to lie, is there are other things that can change the frequency of a variant. Random fluctuations, we call that genetic drift, that can bump variant frequency up and down. You've got to account for that. And then one of the great breakthroughs of ancient genomics has been that the people living in one place are often different from the people who are living at another time. So we see populations moving in, mixing with other populations. So you need to make sure that what you're thinking is, you know, a change in a variant's frequency is not just the signs of one population replacing another, the higher levels of that variant. So you need to account for these things, ancestry and genetic drift. And, you know, studies have done that in the past and had trouble finding signals. And then this study tries to account for it, but it does it sort of in a different way. And what were they looking for then? I spoke with David Reich, one of the co-authors, and he said it was, you know, a very simple test, stupidly simple, I think was his quote. He says, you know, he looks at this variant in different populations in space and time for each variant across the genome and asks, in groupings of people, is it trending up or is it trending down? They're looking for, like, which way is the wind blowing is the term he used for this variant. And after discounting for genetic drift and how different populations are related to one another, what's left? You know, what are the changes that we think are all blowing in the same direction to continue the metaphor? And this identified, you know, about 470 or so gene variants across the human genome that they think have, you know, very strong evidence that they are either being positively selected or negatively selected. And thousands of more variants where the evidence is a weaker but they're still flagged. So previous studies had identified 20 or 30 gene variants with strong signals for directional selection. We're talking about 400, 500 here, maybe loads more. And what sort of things are these variants involved in? Can they be directly linked to a characteristic? Yeah, absolutely they can. But I think we need to do this with trepidation because, like, I talked about lactase persistence. And, you know, it's really easy to come up with just so stories about why a variant was selected or, you know, why it's useful with lactase persistence. You could say, aha, you know, humans invented dairying and, you know, they needed this gene variant to be able to drink milk without getting sick. But if you look back in time, and this isn't from this study, humans were consuming milk and milk products for thousands of years before this variant became a lot more common. And so we still don't fully understand why lactase persistence evolved. And this is the strongest absolute signal in the human genome, which is, you know, something that this study replicates, of course. But looking more broadly, you know, instead of just picking out just so stories, you know, you see lots of evolution for immunity, which makes sense. We think diseases, epidemics, infections were really a major force in recent human evolution. And if you look at this time period we're capturing, the last 10,000 years in Western Eurasia, this is when hunter-gathering gave way to farming. You know, humans haven't lived in this dense populations around this many animals ever before in history. So it's just like transformative time. So it kind of makes sense that maybe we're going to get more pathogen exposure from these changes. So yeah, immunity was, I think, you know, a big driver of change. You're seeing other things, metabolism, heart function. The study is finding a lot of the greatest hits. The other studies have flagged, but loads more. And it's really just a resource for others to see if they can come up with plausible explanations, maybe do some experiments. And it seems to show then that the human genome is very responsive to whatever is thrown at it, right? 10,000 years isn't really that long in kind of grand scales to adapt to stuff. I mean, here we are, so we must have done so. It is suggesting that there's a lot more change going on in the genomes of Western Eurasians, maybe than previously suspected. At least if you buy the conclusions to this paper, which some people do not fully glom onto, it does suggest a lot of change, a lot of directional selection going on. And of course, you and you spoke to a lot of folk about this. I mean, you foreshadowed it a bit there. What's the sense that you get from other researchers about their thoughts on this work? It was mixed. Let's put it that way. And I think everyone agreed that this is a huge resource. This is going to be important for people studying lots of things, including natural selection. And then people said that there's a lot of findings in this paper that are almost certainly correct and almost certainly examples of directional selection. And I just think maybe that some people took issue with the scale, saying it was as widespread as the authors claimed. And I think people took issue with the new methodology in this paper. They thought that some of the changes for some of the variants could be explained by ancestry as opposed to bona fide natural selection. And then there was another aspect to the paper that I haven't yet gotten into. And it's a little bit similar, but instead of like looking at traits that you can link to individual gene variants like lactase persistence, this is looking at more complex traits. So things that we've been studying for a while now with these genome-wide association studies of contemporary human populations, where you get like thousands, maybe even hundreds of thousands of people, and you find all these variants that have a small effect on a trait. And so they looked for lots of traits just to see whether variants that were under selection in ancient Eurasians, whether they affected these kind of GWAS traits. And this is where we need to be really, really careful, but they found that combinations of variants linked to traits including risk of diabetes, risk of bipolar, risk of schizophrenia were under selection in these ancient Europeans. That's one finding. And it's suggesting that risk of those diseases decreased. What you have to be careful with is not saying that these traits were under selection, depression, schizophrenia, type 2 diabetes, but the variants that are predicting them in modern populations, those things were under selection. Similarly, they found that a set of variants that in modern populations predicts things like educational attainment, score and IQ tests, and household income combinations of variants linked to those were also under selection. But of course, you know, these traits didn't exist in the past times and they're also highly correlated. So we really have no idea what actual trait or traits were under selection, only that the variants that in modern populations can predict these traits were changing. Right. So I'm not saying that these conditions were more common in the past or that ancient Eurasians got smarter or wealthier over time. But also care needs to be taken not to extrapolate this out in other ways too. This is a population and there were of course different populations living in different parts of the world. Yeah. And I think an interesting question that is going to be asked and is already being asked is, are we seeing similar signs of selection in other populations like lactase persistence is one. We know that people elsewhere in the world, including in places in sub-Saharan Africa, have lactase persistence due to a different gene variant. And so people are going to be asking this question, how repeatable is human evolution? And so what's the kind of take home from this then? You and do you get a sense like this paper is out now? What are the things that folk could pick up and take on? If you buy the method that the researchers use, you could agree with them that directional selection is a lot more pervasive than maybe we thought in the past. If you're still more skeptical about the methods that have been used, I think your take home is, here's 10,000 new ancient genomes that can contribute to the question and help you answer it. No one's saying, oh, humans haven't changed at all. It's more just about building a strong case from all the other forms of change in the genome to find these signs of directional selection. It's a really hard nut to crack. And I think people are going to keep hammering at it and looking at some of the things I talked about earlier. Well, let's leave it there. You and Callaway, thank you so much for being with me today. Thank you very much for having me. Nature's You and Callaway there. To read his story, look out for a link in the show notes. And that's all we've got time for this week. Don't forget, you can reach out to us on social media. We're at Nature Podcast. And of course, we're on email too. Podcast at nature.com. I'm Benjamin Thompson. Thanks for listening. Are you really buying a car online on AutoTrader right now? Really? At a playground? Yeah, really. Look at these listings from dealers. Wow, your search can really get that specific. Really? And you just put in your info and boom, car's in your budget. Mom needs a second, honey. You can really have it delivered? Really. Or I can pick it up at the dealership. One sack, sweetie. Mommy's buying a car. I think your kid is walking up the slide. Kyle, again, really? AutoTrader. Buy your car online. Really?
