The longevity field hasn’t been very good at naming things, but one notable exception is “inflammaging”: the low-grade chronic inflammation that correlates with aging. Recognition has been growing that inflammation, the ubiquitous reaction of the immune system to various stressors, is a major driver of many age-related diseases and possibly one of the limiting factors for our species’ maximal lifespan.
Dr. David Furman, who has been studying inflammation for many years at Stanford and the Buck Institute for Research on Aging, might be the best authority to talk to about inflammation and aging. Recently, his team developed an inflammation aging clock that he aims to commercialize. David also leads with a personal example by minimizing environmental exposures that cause inflammation.
Tell me about your journey to where you currently are: a renowned geroscientist working at two top-tier institutions on inflammation in the context of aging.
I’m from Argentina. My journey started very early on when I realized I wanted to create a strong positive impact on humanity and decided to try and go for med school. Seeing how some can benefit from the best medicines and others simply have no access to clean water was shocking. I wanted to change that and help people live better and longer. Then, I had a conversation with my dad, who convinced me to study biology or biochemistry, and that way, I could have a much profound impact than being a physician.
I studied biology and then focused on immunology. This taught me how important inflammation is for fighting viruses and other pathogens. In the early 2000s, we first heard the idea that inflammation and the immune system also participated in age-related diseases. It sounded very surprising at the time. We know that inflammation and the immune system protect us from infectious diseases. We also know that if inflammation or an autoinflammatory condition goes awry, you can develop autoimmune diseases, but the idea that non-communicable diseases of aging, like cardiovascular disease or Alzheimer’s, had an immune root was very appealing to me.
That brought me to Stanford. I was recruited by Mark Davis, who I’m sure at some point will get the Nobel Prize for the discovery of T-cell receptors that recognize viruses and cancer cells. A few years after I joined the Stanford community in 2008 as a postdoc at Mark’s lab, he asked me to lead and be more involved with the Thousand Immunomes project, which was just starting at that time.
I became the director of that project, which looked at the immune system at large. We were doing multi-omics before it was even called multi-omics. We were analyzing a few hundred proteins, whereas today, we’re looking at over 10,000, but the premise is the same: by looking at many parameters in human cohorts, we can learn from humans and then apply those learnings directly to humans, skipping animal models entirely. We know they’re broken, right?
So, by analyzing this massive amount of data, I sort of became a data scientist by brute force. I learned from Rob Tibshirani, Trevor Hastie, and Daphne Koller, who essentially invented machine learning and AI at Stanford. Bridging computational sciences and immunology led to many findings and publications, and all of them had this aging component. It was striking that when looking at the immune system, the strongest signal by far was aging signatures.
That put me in a position to focus more on aging and longevity. In 2019, Eric Verdin recruited me as an Associate Professor and to lead the AI platform at the Buck Institute for Research on Aging.
Everything you’ve just said resonates with me strongly. I’ve also been fascinated with the role of inflammation in aging. I agree that its importance appears to be massive and has probably been overlooked. Can you tell me more about the connection between these two things?
Let me give you the historical perspective. We’ve been studying the immune system with decent technologies for about a hundred years, and we all understand that it protects us from infectious diseases, but the idea that aging is partly due to derangements in the immune system only started around the year 2000. It’s a very recent concept. Only 25 years ago, the first paper was published by Claudio Franceschi, who basically said that inflammation resulting from a number of environmental exposures will accelerate aging rates.
That was absolutely shocking to everybody. As a community, we started looking at the pathways, but at that time, we didn’t know what markers or cells were implicated in aging or age-related diseases. It turns out that inflammation not only affects the molecular hallmarks of aging, it can also drive particular diseases of aging.
Take cancer, for instance. I was shocked to see that if you take cancer cells and deprive the media of interleukin-6, they don’t grow, but if you put IL-6 in the media, they start proliferating like crazy. We now know that cancer – from the very early transformation of cells to metastasis and late stage four cancers – is largely dependent on inflammation.
Then we have the relationship between cardiovascular disease and inflammation. We’ve published several papers in major journals showing that inflammation in older adults is largely associated with a higher risk of heart attacks and other heart-related events, like arterial stiffening and ventricular remodeling.
Depending on the specific protein of the inflammatory process you’re looking at, you’re going to see a different effect. For example, eotaxin is a protein typically elevated in older adults that is associated with neurodegeneration. Who would have thought the immune system could drive Alzheimer’s disease? This is changing the paradigm in Alzheimer’s research, in cancer research, and in every single disease of aging as we understand more about the causal relationships between immune system cells and molecules and derangement at the organ level.
That naturally brings up the idea of intervening early. This seems to be an early type of accumulating damage, so theoretically, if we intervene soon enough, we could have a substantial impact on aging.
That’s very well put. I spent 17 years of my career on the idea of identifying early, preclinical signs of disease – molecular changes that are already happening in the body. Using these molecular and cellular changes, you can predict diseases and mortality rates in people who may not even show any symptoms yet.
If you can identify these changes, you can do more than just intervene; you can intercept these diseases. I like to call it the molecular interception of a disease that is developing very slowly and is not yet clinically observable. The whole premise of what we’re doing is not just predicting or understanding biomarkers. It’s about preventing disease and extending the healthspan of the population by intervening early in its course.
It sounds similar to epigenetic clocks, which work because we accumulate epigenetic damage from very early on. Epigenetic aging starts in the womb. Is that also the case with what we might call “inflammaging” or inflammatory aging? How early do the changes you track begin?
That’s an area of research that hasn’t flourished yet; it’s extremely early, so I can’t substantiate claims with strong science. The reason is a lack of sufficient data sets. The theory, however, is that inflammation may start two or three generations before an individual is born.
The concept of transgenerational epigenetic modifications also applies to inflammation. We’ve seen in some studies that the inflammatory state of new generations is impacted by what past generations have suffered from – things like psychological trauma or famine. These events can impact the epigenetic landscape and make individuals more susceptible to an increased inflammatory load later in life.
To give you an example, if an expectant mother has excessive inflammation – perhaps due to smoking or obesity – this can reflect on the growing fetus. Those children will have a higher risk of developing inflammation-related diseases like autism, early diabetes, cardiovascular disease, and even suicidal thoughts. In 2014, we published a paper in Molecular Psychiatry where we studied 500 individuals with major depression disorder and 500 controls. The inflammation levels were all over the place in those who suffered from the disorder. So yes, this starts very early on, possibly even before conception.
You used your 1000 Immunomes Project to build an aging clock that you believe is superior in some ways to existing clocks. If I understand correctly, it uses just a handful of proteins. Can you tell me more about it?
Yes. We leveraged the 1000 Immunomes dataset, which, at the time we built the clock, had data from over 12 years. We built a deep neural network to analyze the data. Protein networks have a lot of redundancy, and human data is typically very noisy. We addressed this using a specific type of neural network called a deep guided autoencoder, which is very different from what most people use for building clocks.
The beauty of autoencoders is that they can effectively deal with redundancy, non-linearity, and noise: the three main challenges in our data. So, it was the ideal tool. Using it, we predicted a person’s calendar age. What I really like about this clock is that it’s not perfectly accurate at predicting calendar age. That inaccuracy gives you room for biological interpretation as to why some people score so much higher or lower than the rest of the population.
So, this clock is trained on calendar age, not on intrinsic capacity?
You’re probably talking about a different aging clock we have that is trained on intrinsic capacity, from a collaboration with a group in France. The two clocks are interrelated, though. High intrinsic capacity, which is a very positive thing to have, correlates strongly with a low inflammatory age. We can explore that later, but for the inflammatory clock, which we call iAge, we predict chronological age. The model is “guided” because it’s trained on two target variables: the immune protein data you feed the algorithm with, and calendar age.
The output is the closest thing to an “immune age” out there. We then use the residual – the difference between a person’s calendar age and their predicted immune age – to see if the clock has clinical validity, and it does. We could associate an increase in the inflammatory clock with having multiple diseases at once (multimorbidity). We also saw a strong prediction of frailty; if I measure your inflammatory age today, I can predict with high accuracy whether you will become frail seven years from now. We then created a gene expression proxy for this protein clock and validated it in external datasets, like the Framingham Heart Study, where we were able to predict mortality in 2,500 people.
I remember you saying that your immune clock singles out centenarians as having a very different immune profile. I think this is amazing because it suggests that inflammation and immune system exhaustion might be what kill the oldest old and that centenarians are people who can somehow defend against this. What can you tell me about that?
Yes, that was one of our clinical validations. We looked at individuals with extreme longevity: centenarians and supercentenarians. We took a cohort of about 20 individuals from the Bologna area in Italy, all of them 100 years or older, and ran our iAge analysis on them. On average, their inflammatory age was 40 years younger than their calendar age.
There was one super-healthy 105-year-old male who had almost never seen a doctor. His inflammatory age was 25. That’s 80 years below his calendar age. He is an outlier, a really interesting person. What is it about his immune system that allows him such a level of control over inflammation? We don’t know yet. Other studies by Nir Barzilai, for example, show that the immune systems of centenarians are very different from their 80-year-old counterparts. They have peculiar CD4 T cells and a very different microbiome. Perhaps the explanation is that their microbiome and immune cell proportions are just shaped differently, but the fact remains: in supercentenarians and centenarians, their inflammatory age is dramatically lower than their calendar age.
Mimicking the immune system of centenarians can help us compress mortality. However, they also eventually die, and immune exhaustion is emerging as a central cause. By doing something about that, we might even be able to go one step further and extend maximum human lifespan.
That’s a hypothesis, right? We can think of multiple ways to try and push the healthspan of the population to, say, 120 years old, so people can be super healthy and then die quickly. But, I want to stress one thing about the difference between healthspan and lifespan. We all want to live healthier, for sure, but many people think of death as a very negative thing. Why? Because most people lack a humanitarian purpose and achievement in life, something that impacts more than themselves and their families. I think it is pretty scary to die and not leaving a legacy behind. Living is great, and dying shouldn’t be the worst thing that could ever happen to you.
Let’s pivot to something more down-to-earth. You said your inflammation clock gave you clues about what works in terms of diet, exercise, and other interventions.
I’m not sure if you’ve seen an article that is coming out in Business Insider about my experience reducing my own inflammation by modifying my environment and lifestyle. I was at a longevity investors meeting in Switzerland and told a reporter my story. She found it amazing and wanted to write a piece on it.
For 15 years, I’ve been studying the social and lifestyle determinants of inflammation, and I decided to start testing these principles on myself. It significantly changed my family’s life. We moved from the Palo Alto area to an off-the-grid cabin in a small valley called San Gregorio. There, we decided to apply the principles of evolutionary medicine to control inflammation.
The idea comes from a paper I published in Nature Medicine in 2019, which basically states that many environmental and lifestyle factors drive inflammation. I hate to call them “choices,” because someone in Fresno breathing polluted air has no choice. I dislike when people say, “lifestyle choices,” because for 90% of the population, there is no choice.
Anyway, I was guided by the following principle: if you move any species to a new environment that it has not evolved with or adapted to, it will develop inflammation as a natural response. The corollary is that the more distant a person’s life is from our species’ evolutionary experience, the more inflamed that person will be.
The immune system acts a sensory system of your environment, much like hearing capacity or vision; the only difference with classical sensory systems is that the output (inflammation) doesn’t reach your consciousness. You can measure this, and you can start thinking about how to implement this principle in your day-to-day life. It’s everywhere: the workplace, your household, your city. There are external and internal factors, some you can control and some you can’t. It applies to water quality, air quality, the food you eat, and the household products you use.
If your body hasn’t seen something during its two-million-year evolution, you probably shouldn’t be exposed to it, because it will cause inflammation. This applies to plastic containers with phthalates, and the microplastics and nanoplastics we are breathing that accumulate in our organs. Everything converges on inflammation and reactive oxygen species. When you read the literature, it becomes obvious that every one of these insults signals to your body through inflammation. That’s what causes issues in the brain, the heart, and the joints.
So, we made changes. We cut out wheat; humans haven’t been exposed to it for more than 8,000 years. Same with dairy products; we only started domesticating cows a few thousand years ago. Then there are hyper-processed foods. There’s a very long list of things you can start tweaking. And then you have to measure the effect. The problem is that the canonical markers of inflammation mostly work for acute inflammation.
You mean markers such as C-reactive protein (CRP)? It probably doesn’t tell you much about chronic inflammation.
Right, it doesn’t tell you anything. It’s worthless for this purpose. People look at CRP because there’s nothing else, but for predicting cardiovascular disease, its accuracy is about the same as flipping a coin. Paul Ridker built part of his career around CRP and now IL-1β, and he loves the idea of drugging these things.
High-sensitivity CRP is widely used, but a savvy cardiologist will tell you they don’t find it that useful for predicting risk. They use it now for suspected acute infections, of course. But it’s an acute-phase reactant; it goes up, but then it comes back down. IL-1β is similar, sometimes chronic, sometimes acute. Some proteins reflect chronic states, while others just change for a short period.
In terms of other interventions, simply not moving around will increase your inflammation. Your body will interpret a sedentary state as a sign that you’re sick, it’s actually called sickness behavior in psychology.
Basically, inflammation is a lifelong, adaptive reaction that can also be very destructive.
Exactly. It can be. Inflammation is built for repair and protection. Your skin and microbiome interact with inflammation all the time, but it becomes very detrimental if it’s sustained and doesn’t resolve. There’s remarkable work from Charles Serhan at Harvard on the biomarkers of inflammation resolution. That’s a whole other area of research that is super interesting and could be pivotal for finding solutions.
I want to ask you about one of your companies, Edifice Health. Is this how you’re commercializing the iAge clock?
Yes, exactly, and let me give you the high-level answer for why. The system for academic research is not ideal; it’s really broken. Think of the amount of money that goes from the government to academia. It’s incredibly inefficient. Why? Because in the academic setting, there is no incentive or training to start a commercial entity. If you don’t do that, the findings end up in a drawer, in the trash, or just as a publication. They don’t translate to the bedside, to households, to solving people’s problems in the marketplace. Federal money does not equal translation. The pathway is not from bench to bedside; it’s from bench to company to bedside.
It seems like the new agency ARPA-H is taking a different approach, and you have applied for one of their programs, correct?
Exactly, they are following this principle. A program we just applied for, called PROSPR, requires us to have FDA approval and a working commercial entity by year five. It’s very pro-startup; a commercialization strategy is a requirement for this ARPA-H funding. It’s a beautiful, dream program. It’s what everyone should be doing: putting their efforts into early diagnosis or interventions for aging but with a translational lens.