Fourth session of the Longevity Therapeutics summit. For coverage of previous sessions:
Biomarkers for Basic and Translational Research (Ronald Kohanski, Deputy Director, Division of Aging Biology, National Institute on Aging, NIH)
Ron got the morning off to a quick start by framing an important question: Now that we think of aging as a druggable condition, how has that changed the way we think about drug development and research in the field? To answer these question, he touched on several important ongoing research efforts (supported by NIH) that bear on this issue.
Intervention Testing Program: Older mice can be healthier, implying a pivot from healthspan to lifespan
Heteochronic parabiosis (HCPB): the transposition of aging phenotypes when the circulatory systems of young and old animals are connected. This paradigm is being currently being explored in the context of age-related declines in bone density. The effects of HCPB may involve a complex set of factors, which include cells that may or may not engraft in the recipient.
Metformin has shown us that we can influence the rate of aging in humans, at least at the level of survival (we’ll be hearing a lot more about this from Nir Barzilai later today)
Heterogeneity: we know that there is a fast-aging population that requires arly intervention to prevent age-related disease and disability, as well as a slow-aging population with the potential to reveal novel protective mechanisms. This influences the way that we should design intervention studies.
He then turned to biomarkers of aging, which he defined as metrics for evaluating aging as a druggable condition. As an conceptual introduction, he discussed the way in which we use morphology as a proxy for time in examinations of fetal development, and composite metrics as a measure of how successful development has been (e.g., the Apgar score). Next, he outlined other scales used to assess development from birth, through puberty, to adulthood. Notably, while fetal development is very ‘standardized,’ maturation to puberty is heavily influenced by environmental factors—Ron pointed out that this is a good analogy for the kinds of things we’re looking at in terms of the rate of aging.
In the history of looking for biomarkers of aging, we have historically used lifespan as a key parameter, but Ron argued that this is a bad idea, and that it would be preferable to score health rather than death. So, how do we score health at different ages? Several composite measurements have been proposed:
Seven activities of daily living
Fried frailty index
Resilience index — doesn’t quite exist yet, but has been proposed. This would include factors such as return to resting heart rate, recovery of activity or muscle mass after inactivity, weight regain after food depirvation, tolerance of water deprivation, tolerance of cold stress, and distrction
MCD Score (to be used by the TAME trial of metformin): the number of chronic diseases associated with aging. Allows counting the time between when a subject has N diseases and N+1 diseases. This could be applied to humans, lab animals, and companion animals (pets).
An extended form of the Apgar score that is valid throughout the lifespan. Would involve addition of cognitive endpoints.
In light of these ideas, how should we handle the (cellular and molecular) biomarkers of aging?
Know the biology that the molecules report.
Use multiple physiological reference points
Deal with heterogeneity.
Consider bifurcating populations.
Have reference and validation populations.
Test against interventions that holistically improve health.
He closed with a generally positive outlook on the value of animal models in studying human aging, with the qualification that we should always take care to ensure that the models recapitulate all (or at least most) of the relevant phenotypes of age-related disease.
Altered Methylation Landscapes in Biological Aging (Morgan Levine, Assistant Professor, Department of Pathology, Yale School of Medicine)
Dr. Levine’s began as a mathematical demographer, and became fascinated by the question of why some people age more rapidly than others—i.e., the heterogeneity of aging. Chronological age is an imperfect proxy for biological age, and quantification of biological age could provide an endophenotype for basic research and screens, as well as facilitate evluation of geroprotective therapies (interventions aimed at delaying aging). [N.B. that Levine believes that people have more than one biological age, because different systems of the body
What makes a good aging biomarker? It shoud…
Track with chronological time
Be minimally invasive
Better predict mortality/morbidity than chronological age
Respond to interventions or lifestyle changes
Facilitate understanding of basic biology of aging (i.e., no “black boxes”).
The goal is a “cradle to grave” quantification of aging at any time int he life course. Aging is exceedingly complex, and therefore it is unlikely that we will ever develop a single aging biomarker that is equally useful in all contexts. Molecular markers allow us to measure the age of a tissue or cell rather than the whole organism, which is likely to be more valuable than scores that abstract every aspect of an individual’s physiology into a single score.
To start thinking about how to quantify age-related change, Levine described a trajectory of aging in which biological, physiological, and functional aging occur in sequences. Molecular alterations lead to physical dysregulation, which leads to disease and disability that are ultimately responsible for death.
Chronological age has been shown to correspond with distinct changes in DNA methylation at specific CpG sites, yielding so-called “epigenetic clocks”. These clocks are highly accurate, and can predict chronological age within a few years. However, Levine argues that rather than minimizing the residual (making the clocks more accurate), it would be better to capture the “true residual”, which is in turn a reflection of the heterogeneity of aging across individuals.
To capture this idea, Levine developed a “healthspan clock”, published last year. Rather than training their clock on chronological age, they created a phenotypic measure of aging that also included nine other marker. Her new clock, based on DNA methylation measurement in whole blood, predicted healthspan better than two previously published clocks, including the well-known Horvath clock (notably, Levine did her postdoc in the Horvath lab).
Even though the clock was trained in whole blood, it tracks with 35 different tissues, ranging from brain to saliva (a material that can be obtained even more easily than blood). Interestingly, however, while brain age is correlated to blood age, brain appears significantly “younger” than other tissues, perhaps implying that maintenance mechanisms are more active in the central nervous system.
So, what are the differences between the existing clocks? They’re trained to predict the same thing, so why do they perform differently? The residuals are correlated, but imperfectly. Strikingly, the models share very few CpG loci, so it seems that they are detecting different signals.
Current work in the Levine group aims to connect the physiological hallmarks of aging to epigenetic changes in DNA methylation.
Conclusions: The new clock, trained on an intermediated marker of healthspan, may be more useful as a measure of aging. Existing epigenetic clocks exhibit diverse associated with age, tissue and outcomes. All major clocks have similar transcriptional signals.
Aubrey de Grey (CSO, SENS Foundation & VP of New Technology Discovery, AgeX Therapeutics)
Aubrey started by saying that he was giving a talk other than the one he’d intended. Specifically, he discussed a recent paper arguing that Jeanne Calment, arguably the world’s most famous supercentenarian, died in 1934, and her daughter assumed her identity to avoid estate taxes. If this is true, then “Jeanne Calment” (née Yvonne) lived to be only 105, not 122. The author of the study argues that “the phenomenon of Jeanne Calment may prove to be an instructive example of the uncertaintly of seemingly well-established facts.”
Why is this important? It’s history, after all, not biology. Partly, it’s because individuals like Calment are widely known around the world, so changes in knowledge influence how the field is perceived by the public. It is vital, in de Grey’s view, that the field evince its ability to reason with uncertainty (and avoid reasoning from data that are in any way inaccurate).
Therapies for Thymic Regeneration and Cholesterol Catabolism (Reason, Co-Founder & CEO, Repair Biotechnologies)
As we age, our thymus gland shrinks, leading to decreased naive T cell production and a compromised immune system. Repair Biotechnologies is approaching this problem by devloping a way to upregulate the gene FOXN1, a transcription factor that regulates thymic activity and production of naive T cells.
The goal is a therapy, delivered via intravenous injection, that will enlarge the thymus in an adult. What is the best way to delivery such a drug to the thymus? As a first approach, Repair Biotechnologies is using AAV, a reliable viral platform that has already been shown to be safe in early-phase clinical trials.
There are several options in terms of clinical indications:
Cancer risk and relapse: Cancers can be recognized and eliminated by the patient’s own immune system, but emerge as a clinical problem when immunosurveillance fails. An increased rate of neive T cell production couldcompensate for failing immunosurveillance and bolster the effects of cancer treatments.
HIV non-responders: HIV infection exhausts the immune system, decreasing the number and functional capacity of T cells. A regenerated thymus could prevent this outcome, especially in patients who do not respond to standard anti-retrovirals.
Native T-cell shortage: the thymus atrophies with age, causing a decline in T-cell production, reducing protection against new pathogens, cancers, and senescent cells. Regneration of the thymus could restore youthful levels of T-cell production.
For a variety of reasons, cancer is the most likly choice: proof of concept and disease model data are available and the financials are clear.
Another target for Repair Biotechnologies is atherosclerosis, which Reason argues is not a disease of cholesterol but of macrophages—a “macrophage death forward-feedback phenomenon”. Current therapies decrease the level of overall cholesterol, which decreases the level of oxidized cholesterol that creates plaques, which in turn reduces the level of macrophage death—in other words, it simply slows the progression of disease.
By contrast, the damage repair approach seeks to eliminate plaques that already exist. Repair has licensed a technology that gives macrophages the ability to catabolize cholesterol, obviating the necessity for them to hand off plaque cholesterol to lipoprotein particles in the blood.
Tales of the molecular garbage man (Kelsey Moody, CEO, Antoxerene/Ichor Therapeutics)
Moody, who also gave a workshop in a breakout session before the conference began, devoted his talk to ongoing work at LysoClear, one of the companies in the Ichor Therapeutics ecosystem. LysoClear has set its sights on age-related macular degeneration (AMD), a degenerative condition that is one of the primary causes of age-related vision loss in the developed world.
AMD is caused by accumulation of lipofuscins in retinal pigmented epithelial (RPE) cells. When accumulation outstrips the cell’s ability to deal with lipofuscin, the compounds are deposited outside the cell in bodies called drusen, which interfere with retinal light detection. As the disease progresses, the RPE cells themselves “choke” on the accumulated lipofuscin and die.
LysoClear is seeking to eliminate lipofuscin using techniques similar to the ways that lysosomal storage diseases are already treatd in the clinic. This represents an important unmet medical need, as there are currently no FDA-approved drugs for the “dry” form of AMD, and the therapies for the “wet” form have a variety of efficacy issues.
Early efforts were devoted to establishing proof-of-concept that the company could remove lipofuscin using an exogenous enzyme. They identified a recombinant manganese peroxidase (rMnP) that was capable of degrading all of the lipofuscins present in the aging retina, although the most prominent species (A2E) was the most resistant to degradation. In mice intravitreally injected rMnP is efficiently taken up into RPEs, and efficiently eliminates lipofuscin.