Prof Brian Kennedy: Unanswered Questions in Ageing Research

[00:00] So, you know, you kind of know you're starting to get old when, like, someone asks you frequently to give the last talk at the meeting.

[00:20] And then they say, oh, it's an honorable position. You should take that. And it's even worse when it's your postdoc that asked you to do that. And then they make it even worse by saying, maybe you don't want to talk about data, Brian. Why don't you summarize the field and talk about the unanswered questions.

[00:40] what that says about me, but I'm going to try to do my best, Max. I'm going to kind of follow the lines of Matt and, you know, Vadim a little bit and sort of give a sort of an overview. And some of the unanswered questions I bring up have been discussed during this meeting. And I think that's good.

[01:00] Maybe this is a little bit more of a summary in that sense, but it's fruitful areas, hopefully, to continue discussions. I do research still, and we're really interested in looking at interventions in biomarkers, and you've heard about, especially about biomarkers at this conference, but also about intervention.

[01:20] interventions, and we're really quite interested in how to combine interventions together. I'll have more comments on that in a minute, and also how to personalize interventions. So a lot of the lab is using all kinds of animal models, including humans, which are to address this question. And I think I will make one comment, and I guess

[01:40] this would fall under the where we need to be anyway. We have huge progress in this field and I would put that on four different verticals in the last 10 or 15 years. We have lots of interventions now. I think I agree with Matt that there are a lot more in unique pathways that haven't been found yet.

[02:00] But there are a lot already. There are a lot of diagnostics. These could be clocks, but also standard markers of aging, physical function, cognitive function, et cetera, individual markers. We have a lot of hallmarks. Pretty soon I think we're going to have as many hallmarks as we have proteins, but we're not going to have any.

[02:20] quite there yet, but people are listing 25 and more now in some papers. And we have outcomes which are functional decline outcomes and disease outcomes. I think what we haven't been very good at doing yet in this field, and it's not an easy thing to do, is to sort of connect these verticals. And so I think one of the

[02:40] One of the challenges we're trying to take up is how do we take a particular intervention, I put AKG on here, but it could be anything, figure out which of these clocks are being affected by it and diagnostics, which pathways are most affected and what are the key outcomes that come from that. I still kind of view it as interventions as key.

[03:00] keys and diagnostics as locks and we need to figure out which keys go into which locks. And once those doors are open, what are the physiologic outcomes that emerge? This is not always so easy. It's not that easy in humans to measure hallmarks of aging because you are restricted by what kinds of samples you can collect.

[03:20] And I still think they're not great diagnostics for different hallmarks in blood, for instance. So there's work to be done here. But let's get to the questions rather than talk about that. And I've just listed them in sort of semi-free form because that's how my brain works.

[03:40] So I'll start just right at the top, or bottom, for instance. And I think one of the big questions that I've been struggling with for a while now is, should we approach the problem of aging from a bottom-up strategy or a top-down strategy? And I put this on here. This is like the poor stepchild of the Hallmark's paper.

[04:00] paper. This is the Pillars of Aging paper that was written about the same time. And I was asked to write this, but I wasn't given a choice of the pillars. This came from a conference sponsored by the National Institute of Aging where we talked about each of these topics and aging and then how to put them together into a review.

[04:20] I had mainly was to draw lines between them because I felt already at that time that it's a little bit disingenuous to talk about these different pillars of aging as if they were unique from each other because they're not. They're all interconnected to each other. And I, as you can see, have a lot of fun.

[04:40] can already begin to probably guess at more of a top-down view on aging. But I think that, as Matt said, focusing on the concept of dealing with each pillar or each hallmark separately and imagining putting all that together is going to lead to an immortal organism, I think, is a flawed concept.

[05:00] think most of the interventions fall under this entrainment idea. When you have something that extends lifespan, it pretty much hits all the pillars or all the hallmarks. Now, it may hit some more directly than others, depending on the intervention, but at the end of the day, you can see improvements in pretty much all of them, which is a great way to generate

[05:20] rate papers for your lab. You show that something extends lifespan and then you test its effect on each hallmark and there's some effect. But I'm not sure what that's really telling us in the long run. And you can also apply this concept to biomarkers. There's some people that really think that we need to.

[05:40] need to be as granular as possible, that a systemic biomarker of aging or clock is not that valuable. What we need to know is how each organ is aging. And if you measure that, you'll find that they're aging at different rates and in different people.

[06:00] still tend to believe based on this network concept that there is a systemic process that maintains homeostasis in the body and that's what's keeping us healthy. And so there is still a lot of value to having a systemic measure of aging. And in my view it's when that network breaks down

[06:20] That's when bad things happen and that's when organ-specific clox or tissue-specific clox probably become more valuable. Now, certainly if you have a genetic mutation that causes some severe problem with a particular organ, that will lead to a disease in a young individual. But I think for the majority of the case, if the homeostasis is working, you're largely protected.

[06:40] And I threw this in there because I'm curious how this applies to aging research because there's been a lot of motivation now to move toward very large collaborations that are funded at a very high level. And I've collaborated with a lot of labs. I like collaborating with labs. I'm not against collaboration.

[07:00] But it raises a question of where discoveries come from in aging research and how much we should focus on ideas of individuals that are unique and maybe a bit crazy and how much we should focus on pulling people together to tackle big problems. Obviously, the answer is somewhere in the middle. But I think right now we're trying.

[07:20] We have a lot of us have this concept that we need to all pull together as a team and solve aging. But I still think that some of the big solutions to aging are going to come from somebody we never heard of yet working on a small project in their lab, you know, in Ohio or somewhere. And so it's trying to find the right balance in how you.

[07:40] fund aging research. Of course, in the US they're just not going to find research at all, so I guess that question is not that important at the moment. Another question is that, this is what's on my mind is, is there really a longevity normalizing effects versus longevity extending?

[08:00] effects. We used this concept in Camille's paper and I think you probably, most of you heard Camille talk yesterday about this, showing that there's a huge problem with short-lived controls in animal models and when mice live short, the best you can say about an intervention that extends

[08:20] life span is that it's longevity normalizing. In other words, the mice are living short for some reason, and bringing their life back to normal is something that intervention did, but it doesn't prove that it's life span extending. Whereas if you have interventions that like Rapamycin and others that dramatically push the

[08:40] lifespan over what's considered a reasonable lifespan for the mice is likely to be lifespan extending. So it's an interesting way of differentiating between the general protectors. But is it something that's different? Are we talking about a continuum? Are we talking about distinct phenomenon?

[09:00] look at humans where a lot of interventions I think will affect median life expectancy, are we just normalizing everything? Are we just taking people that are choosing the wrong lifestyle options and bringing them back to their maximum potential with these interventions? Are any of these interventions going to actually extend lifespan?

[09:20] I think that's a question that's open. This question's been discussed for a long time. What calorie restriction? People used to argue about whether calorie restriction was really extending lifespan, because in mice, you put them on this diet that you call normal, but it's actually a high-fat diet. It's not as high-fat as what we call a high-fat diet.

[09:40] diet, but it is a high-fat diet, and then you reduce the calories and they live longer, are you just optimizing the diet or are you extending the lifespan? I still think this question is not fully answered yet. And so we need to, I think, think more deeply about

[10:00] about this and whether there's a distinction. Now I'm not against longevity normalizing. If we can do that for enough people, that's a massive effect on human health. But I'm also quite interested in whether lifespan extension is feasible and to what extent. How do we combine interventions?

[10:20] know this guy, I think, his name is Brian Johnson. And I put this up here because he had this show out called Don't Die. And at first I wasn't that excited about watching it. But I have to say, everyone should watch it for one reason. And that's because watching the facial expressions on Vadim when he gets asked questions in this

[10:40] movie is worth the ticket alone. Vadim, if you ever want to play poker, just let me know. So we have people out there that are combining tens and 50 interventions and thinking they're going to live forever. I don't even know how to do this with

[11:00] two or three interventions in mice. And I think the question is, is there some strategy to do this beyond my bad guesses? Or how much how far can you actually push life span by combining interventions with the interventions we currently know?

[11:20] These are open questions that are not answered. We heard earlier about some modeling that might help predict interventions that work together. I think that kind of thing is very exciting and very apropos to this conference. I'm not against hackers at all.

[11:40] I'm actually for empowerment. I think people that want to educate themselves and make their own choices about their health to a reasonable extent should be allowed to do that. And if Brian Johnson wants to do whatever he wants to do and kills himself, that's his problem. I kind of worry a little bit about it though because when you're out there

[12:00] advertising how to do something to millions of people and the science behind it is not particularly solid, then I think it's more of a concern. But it's interesting what people are doing. I will say that. And also this comes into another question which is how do we figure out which interventions work in which people?

[12:20] Everyone's talking about precision longevity right now. When you go to a clinic, they tell you, not every clinic, but some will tell you, we're going to measure you and give you a specific recommendation for how to maximize your lifespan. I'm not at all convinced that's possible. I think the only way to get to any kind of personalization at the moment is to get to your physician.

[12:40] is to start people on one path, measure how things are changing, and then try to course correct going forward. And eventually, I think if enough visits, you might get to something relatively personalized. But right now, at least in my view, I haven't seen any strategy that allows us to look at someone at baseline and do much in the way of progress.

[13:00] precision.

[13:20] commercially, they're real fluctuations that are happening and I think we're debating what that means and there's not much overlap between different clocks. I do want to spend a little bit of time on one clock and show you a little bit of data related to this in a second. I do also want to point out that these clocks

[13:40] have multiple purposes and we think about them for different uses and I think use cases help dictate which clocks you should use for which purposes. For instance, we like to look at biologic agent populations. So we have a study we haven't published yet with a bunch of collaborators in Singapore looking at...

[14:00] 10,000 Singaporeans of different age and measuring their biologic age. And the data is quite interesting at that population level using four methylation clox. We'd really like to use these cloxes in points in intervention studies though. And measuring an individual's biologic age and knowing what that means.

[14:20] I still think is very challenging. And so at the population level, I'm already convinced the clocks have value. At the individual level, it's questionable at this point. But I think this third line is important too. We'd like to have clocks that are actionable in clinical practice. And right now, when you go to a doctor and say these 14-

[14:40] methylation sites are telling me I'm too old, the doctor is unlikely to know DNA methylation that well and even if they do, they're going to say, so what? What do I do about that? And so I want to tell you about a clock where we try to address this and by we, I mean Jan Gruber and Fong Sheng.

[15:00] with a very minor contribution from my part. And so the idea was to take clinical parameters and build a clock from it, and this was published in Nature Aging a few months ago. And I'll show you a little bit of data on that because Jan didn't show it. Clinical parameters are great because they solve at least one problem.

[15:20] There's reproducibility. When you're measuring HBA1C and you're measuring LDL, there's much more reproducibility in different labs across the world than there are when you're measuring DNA methylation or proteomics or something else. And so I think that was one exciting component of this too. So what's

[15:40] What they did is they took the NHANES dataset, which is demographic data, medical exams, clinical labs. You can see blood pressure, LDL, glucose, and a bunch of other things. Took intake from one year in 1999, I think. And they used a mortality predictor with principal components.

[16:00] They used one year to build the clock and then predict mortality in the next clock. I'm just going to show a few slides on this. And so this is the test case, and so we're looking at survival in 200 months. So you can infer biologic age backwards from mortality risk. That's what we're doing here. This is the survival of the-

[16:20] Haines population in that year. The red are the top 25 in cardiovascular score and the lower 25% in cardiovascular score. And you can see this clock is better at predicting mortality than the ASCVD. And I think more importantly it's better at predicting survival, way better.

[16:40] than the ASCVD score. So there's some ability to predict mortality using clinical parameters. And there's a couple points that I think are most exciting from this clock is you can look at this data set and you can learn some things that are maybe actionable from a regulatory perspective. Running out of time already.

[17:00] Great. So quickly, you can select people that are optimal and that declines with age. Of course, these are people, all the clinical markers are in the reference range. There are also people in this dataset that are requiring treatment but are not getting it because you know what medical, what, medicine.

[17:20] medicines they're taking, so that's fine. So then they have different life expectancies, so the optimal are living longer and the people requiring treatment are dying faster. You can measure their biologic age too and you can break the optimal people down even further on going fast. So the optimal people that are

[17:40] taking no drugs. So these are people that come to the doctor's office and everything looks fine at 65 to 75 and the doctor is saying everything is in the good range. Try to exercise more maybe, but you know you're doing great. You can also look at people that are optimal, but they're taking pills to get there. So their clinical markers look good, but they're taking

[18:00] medications, blood pressure medicine, or something like that. And the interesting thing is they have a lower biologic age by our clock than the people that are not taking drugs, and that's reflected in the mortality. So the people that are taking pills live longer, have lower mortality than the people that are on no drugs. And so I think that one suggestion from this is you need to be more

[18:20] aggressive perhaps about getting people on drugs and getting these markers into the optimal range earlier and not waiting until things are too bad. Second thing might be that these medicines are already longevity drugs, statins and blood pressure drugs, because they're protecting the network, not just

[18:40] blocking progression along a certain disease. I'm going to go fast through this. This is newer data that's coming out, and this clock works as well or better at predicting mortality than any of the other clocks out there, methylation clocks. It also can predict cognitive score, gait speed, things like that.

[19:00] And you can take individuals and look at the principal components and each component means something. So here's somebody with diabetes, 72. They were predicted to be 88. They died five years later. And the two principal components that are driving their elevated age prediction are component 1, which is basically metabolism, and component 31, which is...

[19:20] to smoking, whereas a person that's healthy, they don't have those parameters. They were predicted to be 64 at 72 and live to 91. So you can look at individuals here and they're clinically actionable things you can do from the biologic age prediction because each component is composed of things that are treatable. Okay.

[19:40] I'm going to go back to questions because I'm running out of time. Standardization. Oh, sorry. I just turned off the whole thing. Standardization. How do we, when do we standardize in this field? I think that's an open question.

[20:00] First, you have a research program where everybody's doing research on a topic and there's a whole research space to explore and people are taking all kinds of different strategies. And so when you do this, you're more likely to make the big discoveries because you're exploring more of the research space.

[20:20] But you're less likely to get reproducibility because everybody's doing something different. I'm going to take a little bit of prerogative. I'm sorry. The other option is that everybody comes together and finds consensus and you come to a clinical protocol, for instance, to treat someone, which is where you ultimately want to be when you're treating people. But you'll notice if you get there too soon,

[20:40] You're doing you're only exploring a very small area of research space and so you get great reproducibility But not a lot of exploration and so the question is where are we in research? Where are we in research for biomarkers? Where are we for clinical intervention studies mouse aging studies? I think that's a very open question now that doesn't

[21:00] mean I'm against things. The aging biomarker consortium is great. Trying to pull together people and decide what aging biomarkers are, I think that's wonderful. But I also don't want to get too far down this path where we decide how to do an experiment and we're not letting people do enough exploration and we're missing the big discoveries.

[21:20] And then I also want to talk, I want to go through this slide. Longevity clinics are doing everything you can imagine and every way you can imagine under all kinds of different regulatory rules and there are no set approaches for how to do this. I'm still supportive of what longevity clinics are doing generally. I think

[21:40] it's important to work with them to try to collect data, to try to help with what approaches they're taking and what biomarkers they're using. So I'm very supportive of longevity clinics, but it's an interesting perspective right now because it's research done in a clinic.

[22:00] That's a very fascinating thing to think about what we're going to do with. And so I just want to close with one thing, a couple things. One is that a lot of funding has gone into aging research lately, but it's really in a translational arm. How do we push health spend and lifespan? How do we take what we know and apply it? And that's great.

[22:20] But I think what we can't forget is that it's still going to take discovery science to get from here to here, which is what I think a lot of us really want to do. Some of the biggest questions in biology, can we stop aging? Can we extend maximum lifespan? Can we reverse aging? That takes discovery science, and I don't know anywhere that's really focusing.

[22:40] on funding discovery. Discovery-based science is going down because the funding is going down for it. And we need to do something about that. And finally, I'll close with this. This is the first JIRA physics meeting. What are we missing? Should we have quantum mechanics more in this?

[23:00] conference. Are there other areas of physics that we should interface with gerontology and aging research? And so you can email me or Max or Jan or anyone else and we'll consider that as we go forward because this is really about how to build a community

[23:20] We don't want to miss a big part of the community. So with that, thank you and I'm sorry I ran over.

[23:40] Last question you asked is really deep and important for me because during all two days of the talks we have during these two days and posters I looked like, I really didn't see.

[24:00] Any posters or reports which precisely discuss the problem of extracellular matrix aging and in general protein deterioration dynamics and so on, which occurs in extracellular matrix.

[24:20] It's a kind of trap question here. How do you think just your intuition, aging is mostly intracellular process or extra out of cellular process? I think I'll avoid that trap, but I will say this.

[24:40] We've ignored the extracellular matrix, I think, and that's a very dangerous thing to do. I think a lot of aging probably is happening there. These are long-term molecules that may contain damage for a very long time. There's already a lot of evidence on, like, hyaluronic acid from Vera's lab and naked mole rats.

[25:00] a lot of other work, but it hasn't been a field that has gotten much attention and I think it should have more prominence in aging meetings everywhere, but probably also in meetings like this too. One of the things you can do is you have people that you think would be really good, maybe you're the person for this, but suggest that because I think that we

[25:20] We're probably missing names that could be speaking at these conferences too. Yeah, Martin. I have a sort of specific question about the sort of the data on the population data with the treatment and no treatment. Could you, or the polypil and the no medicine, could you?

[25:40] could you differentiate which medicines are, you know, driving this and which medicines are probably bad? I'm not the, Jan and Feng Shun could answer that better than I do, but yes, when you look at individual drugs, and there's not that many in NHANES, so we don't have a lot of options there to look at.

[26:00] But when you look, the ones we've looked at, a lot of them have very small effects. So I think that when you put that together, you see a bigger effect because you've got treated people that are optimized versus untreated people that are more or less okay. You see a big effect, but it doesn't.

[26:20] as far as I've seen, it's not one drug. It's like pretty much all of them have some benefit that we've looked at.