In longevity science, interventions often gain attention long before they deserve confidence.
Typically what happens is something like this: a new biological mechanism is identified that appears relevant to aging. A rodent study shows benefit, perhaps even an increase in lifespan. A biomarker shifts in what looks like the right direction. Gradually a narrative forms around the idea that this new intervention might improve longevity.
Many people then become enthusiastic and begin acting on it.
What is usually missing is not intelligence or data, but a disciplined attempt to follow the entire causal chain from intervention to meaningful human effect, and to ask where that chain is most likely to fracture.
One of the difficulties in studying aging is that it is not a process that unfolds over weeks. Aging reflects gradual changes in damage accumulation, repair capacity, and system regulation that develop over decades (I discussed the underlying biology of aging in more detail here). Any intervention that genuinely modifies that trajectory will tend to produce effects that are subtle and slow.
This creates a predictable bias. Interventions that produce noticeable short-term effects, such as improvements in energy, recovery, inflammation, or metabolic markers, often act on adaptive physiology rather than on the underlying processes that determine lifespan. Such interventions may still be useful. But they are not necessarily modifying aging itself.
At the same time, both harms and benefits in aging biology can take years to become visible. An intervention that slightly reduces structural damage accumulation may show no detectable benefit in the short term, while another that provokes strong physiological responses may appear impressive long before long-term costs become apparent.
This becomes particularly problematic because many interventions are evaluated using short-term markers: biomarkers changes, physiological responses, or subjective improvements in how someone feels. These signals are often treated as evidence that something works, or that it does not.
But unless a biomarker is known to be causally connected to the aging process, its short-term movement tells us very little about long-term outcomes. A change in a marker is only meaningful when we understand how that marker actually participates in the biological processes that drive aging.
Rather than starting with theory alone, it is often more useful to examine specific interventions and follow them all the way through the biological and translational constraints they must survive.
Apigenin attracted attention because it inhibits CD38 in laboratory systems. CD38 degrades NAD⁺, NAD⁺ declines with age, and restoring NAD⁺ has produced metabolic and functional benefits in animal models.
The conceptual chain therefore looks straightforward:
Ingest apigenin → inhibit CD38 → preserve NAD⁺ → improve aging biology.
For this to work in humans, several conditions would have to be met simultaneously. The compound would need to survive digestion and metabolism, reach the systemic circulation in sufficient amounts, enter relevant tissues, accumulate to concentrations capable of inhibiting CD38 inside cells, and maintain that effect over time.
This causal chain breaks at exposure.
Apigenin is extensively metabolized in the gut and liver. Only small fractions reach the systemic circulation in intact form, and measured plasma concentrations are far below those required to produce meaningful CD38 inhibition in vitro. Dietary intake, even from foods relatively rich in apigenin, is orders of magnitude lower still.
The biology itself is not incorrect. If apigenin could be present in sufficiently high concentrations inside cells, it is plausible that CD38 activity could be reduced and NAD⁺ levels preserved.
The problem is that the concentrations required to make this mechanism matter are simply not achievable under real-world conditions.
This is a recurring pattern in longevity interventions: a compelling mechanistic insight that cannot be translated into viable pharmacology.
What would change my view would be direct evidence that achievable human dosing produces sustained increases in tissue NAD⁺ of meaningful magnitude, enough to suggest that the CD38 inhibition pathway is actually operating in vivo.
Mesenchymal stem cells (MSCs) are frequently described as systemic rejuvenation tools, based on their ability to differentiate into multiple cell types. The implied model is that these cells can function as replacement units for aging tissues.
In this view, MSCs are introduced into the body as a kind of repair system: new cells that can move to areas where tissues are damaged or aging, replace dysfunctional cells, and restore tissue function. The idea is that by repopulating aging tissues with new cells, the organism can regain some of the lost function associated with aging.
For this to occur, several things would have to happen in sequence. After infusion or injection, the cells would need to distribute widely throughout the body through the circulation. They would need to reach relevant tissues in sufficient numbers, engraft within those tissues, survive long term, integrate functionally into the tissue architecture, and ultimately replace damaged or aging cellular populations.
This is not what typically happens.
After intravenous administration, most MSCs become physically trapped in lung capillaries because of their size. Although cell size varies, many MSCs are too large to pass efficiently through the small capillaries of the lungs, and as a result a large proportion of the injected cells never reach systemic circulation.
Even among the small fraction of cells that do escape the lungs and reach other tissues, survival is usually short-lived. Many of these cells disappear within days.
Despite these limitations, MSC injections can still produce measurable biological effects. Increasing evidence suggests that these effects arise primarily through paracrine signaling, the release of molecular signals that influence inflammation, immune responses, and tissue repair, rather than through durable replacement of damaged cells.
One important component of this signaling appears to be the release of exosomes, tiny extracellular vesicles that can travel through tissues and deliver molecular cargo to other cells. These vesicles can influence cellular behavior even though the original stem cells themselves may not survive long or integrate into the tissue.
This mechanism is fundamentally different from the idea of systemic rejuvenation. It resembles a short-lived biological signaling event more than the regeneration of tissues through cellular replacement.
Additional complications follow. Cells derived from older individuals may already exhibit age-related dysfunction, weakening the premise of autologous therapies in which a person’s own cells are expanded and reinfused. In principle such cells might require rejuvenation ex vivo before being returned to the body, but this is generally not part of standard MSC treatment protocols.
Durable engraftment across multiple organs remains largely unobserved.
MSCs may prove useful in specific therapeutic contexts, particularly where temporary immunomodulation or tissue signaling is beneficial. The broader narrative of organism-wide rejuvenation, however, lacks a causal structure capable of sustaining it. The mechanism observed in practice does not match the mechanism that the intervention is often claimed to operate through.
What would change the picture would be clear evidence that MSCs can engraft widely and durably across tissues and that this engraftment produces sustained multi-system functional restoration in aged organisms.
Among recent candidates, IL-11 inhibition represents a stronger preclinical signal. In mouse studies, both genetic deletion and late-life pharmacological inhibition of IL-11 extended lifespan and improved pathology across multiple organs. (1)
Importantly, beneficial effects were observed even when treatment began in older animals. This is notable because it more closely resembles real clinical use: interventions targeting aging are typically used by middle-aged or older individuals rather than started early in life.
This is the kind of evidence that makes a new intervention biologically interesting.
However, translation from animal models to humans is rarely straightforward.
In the mouse study where IL-11 inhibition extended lifespan, some of the metabolic improvements observed in the treated animals appeared to involve increased thermogenic activity in adipose tissue. Mice possess substantial thermogenic adipose tissue and can increase energy expenditure through this pathway. Humans, however, have far less thermogenic adipose tissue and a much lower capacity for this type of metabolic thermogenesis. Even if the mechanism exists in humans, its overall physiological impact is likely to be smaller.
Another observation from the same study involves cancer. Part of the lifespan extension in the IL-11 inhibition mice appeared to result from a reduction in cancer incidence. While cancer is certainly an important determinant of lifespan in humans, it plays an even larger role in mouse mortality. As a result, interventions that reduce cancer burden in mice can produce substantial increases in lifespan without necessarily slowing the underlying rate of aging.
In such cases, the survival curves may show lifespan extension even though the intervention is primarily altering cancer risk rather than modifying the fundamental aging process.
This means that part of the lifespan extension observed in the IL-11 mouse study may not translate proportionally to humans.
There is also the question of overlap with other longevity pathways. IL-11 signaling interacts with biological systems that intersect with pathways already targeted by interventions such as caloric restriction or mTOR inhibition. For individuals already using interventions that influence mTOR signaling, the remaining independent effect of IL-11 inhibition may be smaller.
None of this means the finding is wrong. It just means the expected magnitude may be smaller. The relevant question is not whether the biology is real, but how much of that biology survives the transition from controlled animal models to the far more heterogeneous process of human aging.
What would increase confidence would be replication across multiple mammalian systems and evidence that IL-11 inhibition produces additive benefits when combined with established longevity interventions.
A recurring issue in longevity research is the difference between detecting a biological signal and estimating its magnitude.
Many interventions genuinely influence pathways related to aging. The problem is not whether the signal exists, but how large its effect remains once the constraints of a long-lived organism are taken into account. Mechanisms that produce clear effects in mice can shrink substantially when translated into humans with different physiology and much longer lifespans.
Hyperbaric oxygen therapy (HBOT) has legitimate clinical uses, including for wound healing and recovery from ischemic injury. Reports of increased leukocyte telomere length following treatment, most notably in a study by Efrati and colleagues, generated interest in its possible relevance to aging. (2)
However, telomere measurements in circulating immune cells can shift for reasons that have nothing to do with true telomere elongation. Changes in the average telomere length of leukocytes may simply reflect shifts in which immune cell populations are present in the blood at a given time. Cells can move from tissues into circulation or leave the bloodstream and migrate into tissues, altering the composition of the circulating cell pool. Such redistribution can make the average telomere length appear longer without any individual cell actually becoming younger.
The larger issue concerns oxygen exposure itself. HBOT exposes tissues to supraphysiologic oxygen tensions. Oxygen is essential for life, but it is also highly reactive. Organisms have evolved extensive antioxidant defenses to cope with the levels of oxygen normally present in tissues.
Hyperbaric oxygen therapy temporarily raises oxygen concentrations to levels far above those normally encountered in the body, including in compartments where oxygen is typically much lower. Evolution has optimized biological defenses around normal physiological ranges, not these extreme exposures.
Short-term adaptive responses to such exposures may be beneficial. But beneficial short-term responses do not necessarily imply favorable long-term effects.
Known risks illustrate this point. Repeated hyperbaric oxygen exposure is known to increase the risk of cataract formation, likely because high oxygen levels overwhelm antioxidant defenses in the eye. This provides a concrete example of how supraphysiologic oxygen exposure can produce cumulative oxidative damage.
The broader question is whether repeated HBOT over many years produces net benefit, neutrality, or harm with respect to aging. This question remains unresolved because the relevant time scale has barely been studied. Some forms of oxidative damage accumulate slowly and may not become detectable until long-lived biological structures have been affected over many years or decades.
Given the reactivity of oxygen and the long timescales over which structural damage can accumulate, short-duration studies cannot meaningfully resolve this question.
This is not implausible biology. It is insufficient observational period.
What would clarify the issue would be long-duration animal studies or multi-year human data capable of detecting cumulative effects across biologically relevant time scales.
Each intervention above fails for a different reason. Yet the underlying pattern is remarkably consistent.
The difficulty is rarely determining whether a mechanism exists. Most of them do. The difficulty is that between a molecular mechanism and an effect on lifespan, an intervention must pass through several translation filters.
Across the examples above, interventions can succeed:
at the molecular level yet fail to reach or maintain effective concentrations in living tissues,
at altering circulating biomarkers without changing the underlying damage state,
as transient physiological signals without producing durable structural change,
in short-lived model organisms while producing smaller or different effects in longer-lived organisms with different physiology,
at modifying cellular regulation while leaving long-lived extracellular damage largely untouched,
at the correct biological target but with an exposure pattern too brief or too infrequent to meaningfully influence long-term damage accumulation.
These are not unusual complications. They are the typical points where promising mechanisms lose their impact when translated into real biological systems.
Recognizing these filters shifts the question from “Does this mechanism work?” to the more predictive one:
Can this effect propagate across biological scale, time, and structure without being dissipated?
For those allocating resources, designing interventions, or advising others in this space, the relevant question is rarely whether a mechanism is interesting. Many mechanisms are interesting. The question is whether the full chain from intervention to sustained human effect remains intact when examined under realistic biological and translational constraints.
However, decision-making in this field occurs under conditions of deep uncertainty. For individuals, particularly older adults, the choice is often not between a proven intervention and an unproven one. It is between experimentation and doing nothing at all.
In such cases, experimentation can still be rational. Even when the probability of success is low, individuals may choose to act because the alternative, continued aging without intervention, carries its own certain risk.
The appropriate posture therefore depends on context. When allocating capital or designing therapies intended for large populations, caution and strong evidence thresholds are essential. When individuals consider personal experimentation, the calculus can be different.
A disciplined approach is not about rejecting new ideas. It is about understanding where the causal chain is most likely to fail before acting.
Evidence capable of meaningfully increasing confidence in longevity interventions tends to share several characteristics.
First, achievable exposures must reproduce the proposed biological mechanisms in living human tissues, not only in isolated laboratory systems.
Second, interventions should produce biological effects that influence processes known to contribute to aging. In some cases, the intervention itself may be temporary while the downstream consequences are long-lasting. For example, a treatment that lowers glucose for a limited period could still reduce glycation damage accumulated during that time, producing effects that persist long after the intervention stops.
Third, results should replicate across organisms or systems that differ meaningfully in lifespan and physiology.
Finally, evidence should extend beyond simple biomarker movement and demonstrate that the intervention meaningfully influences biological processes involved in aging. In some cases this may appear as improved function or increased survival. In others, the effect may be subtler, such as measurable slowing of damage accumulation, with functional consequences becoming visible only much later.
Longevity interventions should therefore not be judged merely by whether they produce biological change, but by whether those changes survive the transition from mechanisms to real human biology.
Until then, most ideas remain exactly that: promising, but unproven.
References
1. Widjaja AA, Lim WW, Viswanathan S, et al. Inhibition of IL-11 signalling extends mammalian healthspan and lifespan. Nature. 2024;632(8023):157-165. doi:10.1038/s41586-024-07701-9
2. Hachmo Y, Hadanny A, Abu Hamed R, et al. Hyperbaric oxygen therapy increases telomere length and decreases immunosenescence in isolated blood cells: a prospective trial. Aging (Albany NY). 2020;12(22):22445-22456. doi:10.18632/aging.202188