Prostate cancer prediction and bonobo culture
This month on the eLife podcast, artificial intelligence reveals a better test for prostate cancer, is the brain stuffed with neuronal stem cells, bonobos with cultural preferences, and why some insects play “follow my leader”...
In this episode
00:37 - AI predicts prostate cancer severity
AI predicts prostate cancer severity
Georgina Cosma, Loughborough University & Graham Pockley, Nottingham Trent University
There’s an old saying in medicine that, if you live to 80 then 80% of you will get prostate cancer, assuming you’re a man of course. But that doesn’t mean that 80% of men will die of prostate cancer. Some will, but the majority will die of something else completely different. In other words, they don’t need aggressive - potentially quality-of-life-changing prostate cancer treatment. And the problem is that when we find prostate cancer in a person, at the moment we don’t know who does and who doesn’t need that intervention. But that might be about to change. Chris Smith spoke to Graham Pockley and Georgina Cosma, who have used a machine learning algorithm to identify a set of markers present on natural killer cells present in the blood that can much more accurately detect who actually has prostate cancer, and the likely aggressiveness of their disease…
Graham - Prostate cancer affects around 45,000 men each year in the UK. And around 11 or 12,000 men die of the disease. One of the big problems in trying to manage patients with prostate cancer is having a very clear understanding of whether they have the disease in the first place, because the current tests don't work very well. And if they do have the disease, how sort of serious is that disease? Do they need treatment or can they just be left and watched? So the key problem we were trying to solve is to improve the diagnosis of prostate cancer using a simple sort of blood test, rather than having to take tissue out of patients and looking at it under the microscope.
Chris - We do have some blood tests you can use to diagnose prostate cancer or at least monitor it. So what's wrong with that?
Graham - Yes. The most common test is something called prostate specific antigen - PSA test. So this is a blood test that measures the levels of a certain molecule in the blood. The problem with the prostate specific, the PSA test, is the fact that in some cases, the level is normal in men who have prostate cancer. In other cases, it can be raised in men who don't have prostate cancer. So there's a lot of false positives and false negatives, and that causes a problem because it doesn't allow that particular test to be used for screening a large number of men because of the false results it gives.
Chris - And what have you done instead to try to work out a way around that problem?
Graham - The key thing is we still wanted to go with a blood test because that's obviously the easiest thing to do. So we decided to focus on the patient's immune system. As well as protecting people against infection the immune system protects people against cancer and the immune system can recognise the presence of cancer. So we thought that if we took a picture of the immune system in the blood, can that be used to tell whether a man has got prostate cancer? And if he does have prostate cancer, how serious it is.
Chris - Georgina, how did you actually go about doing this?
Georgina - So for this study, we collected and examined the natural killer cells of 72 patients. Amongst these patients, 31 of them had prostate cancer and 41 were healthy patients. So these biological data were then used to produce computer models that can detect the presence of prostate cancer and its severity.
Graham - For the test what we did is we took blood samples and we took immune cells from those patients. And we looked at how they appeared by putting them through an analysis. And so we could identify a whole range of different sorts of characteristics or features of these white blood cells, these natural killer cells. And then we wanted to assess whether there are any differences in the profile of those features or those characteristics between men who did have prostate cancer and men who didn't.
Chris - And that's where you come in Georgina.
Georgina - Yeah. So we initially had a set of 32 biomarkers and the aim was to find the subset of biomarkers that can be used to predict presence and severity of prostate cancer. We ended using a technique, which is a computation optimisation approach, to identify the set of biomarkers that would make good fingerprints for predicting prostate cancer. The fingerprints that we found was then used to build a machine learning model.
Graham - In essence then you know that you've got these features that you can look for that are there, but you don't know which combination are going to be the strongest and most powerful predictors of who's really got disease or not. So having identified the ones you think are going to be the right choices, that's where you then move into building a model so that you can then take this forward and test it against more data effectively to see how good it is, presumably
Georgina - Yes, precisely. So after selecting the best candidate set of biomarkers, we developed machine learning based tools. This resulted in a prediction model that was 12.5% more accurate than the PSA test that's used in clinical practice in detecting prostate cancer. We also developed a second detection tool, which was 99% accurate in predicting the risk of disease in patients with prostate cancer.
Chris - So Graham, basically, you now have a test which is potentially a lot more accurate than just PSA in isolation.
Graham - Yes, that's right. I think generally speaking, you know, the PSA test is only accurate in about 3 out of 10 cases. And so there's lots of men who have a PSA test who are told that they have prostate cancer when they don't. The other issue is that in the vast majority of cases, men die with prostate cancer rather than of it. So it's not in the majority of cases, a life limiting disease, but people are then sort of labelled as having prostate cancer and have to live with that for the rest of their lives. What this test allows us to do is sort of identify the presence of prostate cancer in the first case, but more arguably more importantly is we can categorise that, or we can say whether that disease is something to worry about or something not to worry about. And that allows the clinicians to focus on treating the more aggressive disease.
Chris - And Georgina, presumably you have road tested this by taking people with an unknown diagnosis and then got the predictions from your model and then followed up with a gold standard, perhaps a biopsy or something from those people, to know that you're actually producing accurate predictions.
Georgina - Yes. We also carried out separate tests using a set of records which were not used during the train and test process. So it's like a mini clinical trial basically where we took the finished machine learning model and we input about, say 10 test records, and then looked at the outputs of the model and compared that to their gold standard, which was the biopsy results that we had.
Chris - And critically people are often in a quandary when they're given a diagnosis of prostate cancer as to whether or not they should seek active intervention or a more watch and wait type strategy. Does it help them to make that decision?
Graham - Yeah, absolutely. I think that's the fundamental finding of the project. It allows that decision to be made far more accurately. The approach would reduce the need for about 70% of biopsies, which is quite an unpleasant experience. 1 in 20 men who undergo that procedure get some form of infection. So what it would really allow the clinical team to do is to tell a man, you have a very low grade prostate cancer, but it's absolutely nothing to worry about. There's no need to treat it. Or say you have prostate cancer, it's something we're worried about, and we do need to take it further and then further investigations. But it would remove unnecessary investigations in the large proportion of men. And at the moment that's just not possible.
09:10 - Astrocytes act as neuronal stem cells
Astrocytes act as neuronal stem cells
Jens Magnusson, Stanford University
In the late 1990s, scientists turned one of the central dogmas of neuroscience on its head: they showed that - even in aged adults - new nerve cells were still being born in certain brain areas. Previously we thought that in higher animals like us, neurogenesis stopped shortly after birth. Now scientists have made an even more extraordinary finding: that astrocytes, the non-neuronal supporting cells in the brain, can respond to injuries by transforming themselves into cell types resembling neural stem cells and produce new neurones. It’s early days, but it could have huge therapeutic potential. Chris Smith heard the story from Jens Magnusson…
Jens - The brain is very bad at replacing dead neurons in response to injuries. We had found previously that a very abundant type of support cell in the brain, which are called astrocytes, can activate a neuron-producing capacity in response to certain injuries in mice. But not all astrocytes appeared to have this ability. And we wanted to study the mechanism by which some astrocytes produce neurons, because we think that this might in the long term, enable us to recruit astrocytes as a sort of reservoir for new neurons that could be used therapeutically.
Chris - Traditionally, we think of astrocytes as these supporting cells, but if you injure the brain, you tend to find that they grow a lot and they produce almost like a scar. So that seems to be a little bit contradictory to what you're saying about them now being able to produce nerve cells instead. So are they different sorts of astrocytes that are doing this then?
Jens - We think that astrocytes have different programs so to say that they can turn on. It's true after an injury astrocytes can produce a scar. We found that after stroke injury astrocytes can also generate new neurons. And we wanted to understand the molecular program that creates this response.
Chris - How did you do the study then, talk us through what the method was and then we can perhaps unpick what those results are telling you.
Jens - We used a method called single cell RNA sequencing. So that's a way to read the entire gene expression profile of individual cells. Now, this is informative because cell types and cell states can be identified by the combinations of genes that they express. So we first triggered neuron production by some astrocytes in the mouse brain. And then we looked at their gene expression profiles as they initiated neuron production in mice. We found two things that are particularly interesting. The first thing was that as astrocytes initiated this neuron producing ability, they first underwent changes that made them extremely similar to neural stem cells. So there are neural stem cells in some parts of the brain that continuously make new neurons. And our astrocytes became very similar to them. The second thing that we found was that even though only a minority of astrocytes went on to generate neurons, all astrocytes appeared to initiate this neuron producing program, but then halted their development halfway. And we found that we could inject a type of growth stimulating protein into the brains of these mice and this helped some of the halted cells to resume their neuron production.
Chris - Does this therefore have a therapeutic potential in the sense that you could go into a sector of the brain, which is subject to some kind of trauma, some injury where there's perhaps been some selective vulnerability among a group of neurons where there's been loss, and stimulate the local astrocyte population to differentiate back into and repopulate that missing nerve compartment?
Jens - We hope that in the long term, this is exactly what we're going to be able to do. Now we're far from being there now, but if astrocytes can be viewed as dormant neural stem cells, we think this is really exciting because the capacity of the brain to replace dead neurons is extremely low. And if the brain turns out to be full of dormant neural stem cells, then if we can figure out how to unleash their potential, then that might represent a potential way for improving brain repair after injury and disease.
Chris - What about the phenotype of the nerve cells that you produce in this way? Are they just generic neurons or do you end up with neurons that resemble the native population that correspond to that part of the brain where those astrocytes were when you start starting to stimulate them in this way?
Jens - We don't find evidence that the astrocytes produce neurons that are of the same subtypes of the region that they're in. Astrocytes tend to generate interneurons that are not necessarily abundant in the tissue where they're generated. So we think that in order for this to be useful, therapeutically in the future, other bioengineering approaches are going to be necessary in order to steer the astrocytes towards certain types of neurons that are needed.
Chris - Do you know what the survivability though of those newborn nerve cells is? Because when people have studied the generation of new neurons in brains, cause that does go on in the adult brain, doesn't it, there are certain areas where neural stem cells continue to give rise to neurons throughout life, it's not a given that the new daughter nerve cells survive. So what proportion of the astrocyte derived nerve cells survive and how long for?
Jens - It's low, less than 1%. We don't know if this is a shortcoming of this neurogenic process, or if this is a built in mechanism.
Chris - And do you know if you can shift the survivability, are there common routes by which these cells try to die, that it may be possible to arrest?
Jens - Other groups have found that survivability can be increased by treating the brain with certain chemicals. So I think it's possible in principle.
16:13 - Bonobos with cultural preferences
Bonobos with cultural preferences
Liran Samuni, Havard University
If you were to charter a plane from the capital of the Democratic Republic of Congo, fly for 4 hours, then spend 6 hours on a motorbike, and finally travel for several more miles on foot, you would come across a field site where there’s a unique phenomenon taking place: two groups of bonobos that have distinct cultural preferences. Eva Higginbotham heard what they’re up to from Liran Samuni...
Liran - As humans, we are this crazy special, unique species. We are all over the world. We are capable of cooperative behaviours that have no parallel in the animal kingdom. And one of the things that is considered to be uniquely human is our great capacity for culture and cultural differences, which are group-specific behaviours that are acquired through learning. And I think one of the questions is, where in the history, in the evolution of our species, did we develop the tendency for cultural behaviour or cultural differences, traditions? When we study the evolution of humans, one way is through fossils. We go to excavations and we find fossils that tell us a little bit about the history of our species. But when we want to look at specific behaviours, it's very tricky with fossils because behaviour does not fossilise. And in that sense, we turn to some of our relatives, primates, and especially bonobos and chimpanzees who are our closest living relatives. This paper looks at what we call diversity of behaviours and especially one that is emerging between two groups of bonobos that live in the same environment. Each group is specialised on the hunt of different prey. One group hunts antelopes, small antelope called the Diker, and the other one hunts a gliding rodent. The striking thing is that despite the fact that they live just in the same environment, they hang out in the same places, they still show this group identity in terms of what they feed on
Eva - Is that surprising for animals of the same species that are in the same environment to have that difference in behaviour?
Liran - I think it's very unique and special. Yes, because we can imagine that a lot of how behaviours form and come about is in animals reacting to their environment. If the environment is a certain type of environment, then different behaviours can emerge. But here we have the exact same environment and still almost two different solutions to the same environment.
Eva - Antelope are just such different animals from the smaller rodents that the other group hunts, it seems like such a big difference in strategy. Where do you think that comes from between these two groups?
Liran - That's a great question. And it's a bit of a mystery. One idea that we had is that meat is a high quality food, but it's very rare and it's hard to attain. So hunting frequencies are quite low. And when they're managing to catch the animals, they're super excited, really hugging each other, vocalising, and a lot of excitement. One idea that we have of how this could happen that two groups that live in the same area have such different techniques in a way to avoid competition between them. So if we can imagine that there is this high quality food that is hard to access, and it's very rare, it might be advantageous for each group to specialise on a different prey, so to avoid competition or to reduce the competition.
Eva - What might this tell us about the evolution of culture?
Liran - We know by now that community culture, which is knowledge that is transferred from generation to generation, and the culture is shaped through knowledge that is passed on - I think this is something that we don't have great evidence that nonhuman animals have that. However, the basic capacities of culture, which are two different groups, no differences in their environment, but still showing some diversity of behaviours that is evolved as a function of learning, I think this shows that on the prerequisites of culture as it's seen in humans today. So I think especially the fact that both bonobos and chimpanzees show these abilities tells us a lot about how our ancestors were. Another very important reason why we want to study those animals is in terms of conservation. There is knowledge in the world that is in nature and in wildlife and we can look at it almost as a library full of books that no one has ever read. And through research, we are able to get a glimpse at some of those books. And I think with the biodiversity crisis, it's almost like imagining that this library will be burnt and this knowledge will be forever gone. So I think in some ways, studying those animals, we are trying to preserve some of this knowledge and at the same time conserve these incredible animals
21:27 - Cholesterol levels in the Caribbean
Cholesterol levels in the Caribbean
Rodrigo Carillo, Imperial College London and
Medicine is full of examples where we use data collected on one group of people and then apply it more generally across populations, sometimes incorrectly. Some blood pressure lowering medications, for example, we now know work better in certain groups than others, probably for genetic reasons. Blood cholesterol - or more accurately blood lipid levels - is another example. And working in the Caribbean and Latin America, that’s what Rodrigo Carillo has been looking at, as he explains to Chris Smith…
Rodrigo - So we wanted to know what was the most common lipid problem in Latin America and the Caribbean. There are plenty of large scale studies in Europe, US and Asia as well, but in low-middling countries like Latin America and Caribbean, and also in Africa, these sort of quote unquote well-known questions are still unanswered.
Chris - And obviously this is important in terms of a health priority, because if we want to manage disease and anticipate disease risk, and therefore plan ahead appropriately, we need to know these sorts of numbers rather than just assume that what's true in North America or Europe applies also to other countries as well.
Rodrigo - Absolutely. For this kind of thing, we're perhaps still trying to answer the most basic questions. Like how many are we, how many are there? Where are they? There were some smaller studies in some countries, but looking at the region as a whole, we will still need to answer many questions.
Chris - Did the raw material exist for you to study or did you have to literally embark on this from the get go physically drawing blood and looking at people's cholesterol levels to find this out?
Rodrigo - In this work we rely on published data. So we did a systematic review, which in simple terms is looking through every single journal article, scientific papers, and extracting information when they measure lipids or any sort of cholesterol to have information from as many countries as possible.
Chris - So when you amalgamate all of these research studies, how many subjects, how many people do you end up with to consider?
Rodrigo - I think it was about 150,000.
Chris - So big numbers, these, and when you extract the data, are there any obvious trends?
Rodrigo - Low HDL is not very good for your health. And we found out it was the most common trait.
Chris - What about the other risk factors like triglycerides, and also like LDL bad cholesterol that also furs up arteries? What do they show?
Rodrigo - For LDL we also find quite large estimates, quite a large prevalence though it was smaller than for HDL. For triglycerides we also find large numbers and it was interesting to see that triglycerides was a little bit higher in what we call Andean Latin America, which is perhaps Peru, Ecuador, and Valeria. And it could relate with the sorts of diets they have in these countries, which is perhaps most based on carbohydrates.
Chris - To what extent do you think what you're seeing here is just the environment in which people live and perhaps their lifestyle. In other words, I know if for instance, you drink too much alcohol, you depress your HDL good cholesterol. And to what extent is this relevant to a person's ethnicity? Because obviously these areas have also seen a lot of immigration over the years and those people could quite possibly have brought disease risk with them couldn't they? So how do you dissect out if you can, those effects?
Rodrigo - Yeah, it's difficult to do that with this work. Though I think you are right. And lifestyle will have a lot to do with that, but not only lifestyles, but also opportunities to have a better lifestyle, for example, physical activity and also diet. What sort of diet do you have access to? Unfortunately, like you say, you can't differentiate migration between external and internal immigration and sometimes internal immigration coming from rural areas to urban areas - these people do not usually have the highest standards of living. Of course this has implications for the diet they can afford.
Chris - What's your take home messages then from what you've discovered - depressed HDL, some high levels of LDLs and some high levels of triglycerides as well, and some possible associations with diet. What are your take home messages and how should this inform our understanding of heart disease risk in these geographies going forward?
Rodrigo - The take home message is that LDL is important. And many, if not all, clinical guidance target this with pharmacological treatment. And there is lots of evidence of how lipid lowering medications work and do great things for your health. But these medications do not usually work for HDL. So the take home message I think is that we have to work together from a practitioner perspective prescribing medicines, where these are required, but also from the public health perspective, in which we should give people the opportunity to embrace a healthy lifestyle with access to physical activity, with access to adequate diet. And to include both in clinical guidelines, as well as in policies or strategies. Over the last year, there has been a lot of evidence and a lot of work in obesity and in blood pressure, and diabetes, but lipids and cholesterol overall have not been very well paid attention to.
27:27 - Ant and termite tandem runs
Ant and termite tandem runs
Gabriele Valentini, Arizona State University
Ants and termites show a behaviour called a “tandem run”. It’s where two of the insects venture out, one leading, one following. If you just glimpsed them doing this, you’d conclude that it was the same behaviour carried out for the same reason. But actually ants do it multiple times to teach nest-mates the way to new locations, while termites do it just once after mating to set up a new nest. And for the termites, it’s a case of follow my leader; but for the ants, it’s different: the leader determines the direction, but the follower sets the pace so she can learn the way. To discover this, Gabriele Valentini made video recordings to watch the animals running around in petri dishes and then extracted the information about which insect was calling the shots using image tracking software, as he explained to Chris Smith…
Gabriele - This behaviour is actually widespread in many insects like ants and termites. They use a very similar communication mechanism. So we have the follower touching the rear of the leader to communicate its presence, and the leader instead emitting volatile pheromones that can be perceived by the follower. We also know that, in ants, the leader is making sure that the follower is learning a new path for the environment, while in termites they are just navigating an unknown environment. And so the function of the tandem run in termites is simply to ???
Chris - So how did you actually do the study and what emerged as you did this?
Gabriele - The way we approach this question has been by looking at how leader and follower share information with each other. We performed several experiments where we tracked leaders and followers over time and what we did was to look at how the trajectory of the leader and that of the follower are caused by each other, by looking at how different types of information flows between leader and follower.
Chris - Can you just describe the experiments you actually did?
Gabriele - We did experiments with ants and termites. The setup is different between the two species. Ants are looking for a nest. We are actually starting a new immigration and having them explore the environment, look for a new home. And once they found this new home, they will start to recruit there using tandem rounds. And this is the part of the experiment we're very interested in. So we repeat several of these experiments, making this colony immigrate several times. And for each immigration we then recorded the tandem ants and the trajectory of leaders and followers. The situation is a bit different with termites. This is because they are not recruiting to a known position in the environment, but they use this behaviour only after mating. So we had to collect unmated termites, allow them to mate, and after that we put them in a petri dish where there was no nest they could find. We recorded the entire experiment using a video camera placed on top of this arena. And from the video recording, then we were able to use tracking software to extract the trajectory of the leader and followers.
Chris - And once you piece together, all these videos and watched all these ants and termites running, what did you learn?
Gabriele - There are at least two types of information that are being exchanged between leader and follower. One is information about the direction of motion, whether to turn left or right. And another important information is instead about the pace of their run - when to stop and want to resume motion. Although ants and termites can use the same communication mechanism, what we discovered is that in the case of termites, the leader is controlling both the direction of motion and the speed. And this all makes sense because we know they are searching at random and so the input of the follower to that behaviour is likely to be less important. The interesting discovery was for the ants, where we saw that the leader is deciding about the direction of motion, whether to turn left or whether to turn right, while the follower is actually controlling the speed of the run - when to stop and want to resume. And this was a very important finding because you provide the support for a hypothesis behind the tandem behaviour of ants, that is that the follower is acquiring information from the environment in order to guide a future journey for the environment. So in this case, we have that the leader is travelling through the environment, choosing the direction of motion, in order to show this part of the environment to the follower.
Chris - And why do they therefore need to slow down? What's the benefit of the changed rate from the follower?
Gabriele - We haven't proved it yet, but our main assumption, our main hypothesis, is that the follower is looking around the environment, looking for landmarks, that she can memorise and use as an information to guide a future journey. While they are moving they have a relatively higher speed and the follower is very much focused on not to lose the leader. But from time to time, she takes a break. And during this break, the follower detaches from the leader and walks around a few centimetre apart from the leader, usually making a sort of loop, before coming back to the leader, touching and restarting the journey. So these longer pauses are actually a mechanism, but provide the follower with the time to acquire visual information from the environment. And only when the follower has enough confidence in that information, she will then prompt to resume the journey.