Lower-quality, clickbait-hungry media outlets love sensationalist claims, but one does expect better from the public relations department of an internationally-respected research university. And it was an easy jump from the already-overstated “In First, Aging Stopped in Humans” and “treatments can reverse two processes associated with aging and its illnesses” to saying that a treatment “can reverse aging process” — and to then land in a mud-pit of self-parody with “Human ageing reversed in ‘Holy Grail’ study, scientists say.”
The actual findings of a recent study on hyperbaric oxygen treatment (HBOT) were much more limited. Despite some intriguing indicators, the actual impact of HBOT on aging based on this study is entirely unclear, quite plausibly negligible, and in any case objectively less impressive than that of (say) regular exercise, which certainly does not “reverse aging.”
Originally developed to treat divers with decompression sickness or dangerous nitrogen bubbles in their blood after surfacing too quickly, HBOT involves administering 100% oxygen to subjects resting in a chamber where the atmospheric pressure is artificially elevated. This enhances the dissolution of oxygen directly into the plasma (normally, it’s almost entirely carried in the red blood cells), and the dissolved oxygen in turn drives out the nitrogen, while delivering more oxygen to the tissues. Later, it was found that repeatedly subjecting people or experimental animals to HBOT can cause an adaptive response that paradoxically resembles being subjected to inadequate oxygen (hypoxia) — a phenomenon referred to as the hyperoxic-hypoxic paradox.
Under the relatively loose 510(k) regulations for medical devices in the United States (standards that were modified over 2019-2020 to still-unclear effect), HBOT devices are also “cleared” (but not approved) for carbon monoxide poisoning and a surprising range of other indications, including treatment of chronic wounds and necrotizing soft tissue infections, burn or crush injuries, and unexplained sudden sensorineural hearing loss. Regulation of medical devices is similarly inadequate internationally. Some clinicians also administer HBOT to patients with post-traumatic stress disorder (PTSD) and traumatic brain injury (TBI), although the evidence base for these uses is weak. Device manufacturers and clinics sometimes push the line even further, advertising their HBOT devices and services for yet more speculative indications, resulting in FDA warnings to consumers and the occasional reprimand to manufacturers, although enforcement is hampered by a trend in the courts to recognize broad commercial freedom of speech rights.
So what about HBOT for aging?
The Study Setup
The study recruited 35 independent-living men in good functional and cognitive condition (granted the effects of aging — they were age 64 and older, with some in their early 80s) and had them complete a baseline assessment. They lost five participants right there — a problem that gets compounded as the study goes along, as we’ll see. The remaining thirty subjects underwent 90-minute HBOT sessions five days a week over the course of three months, for a total of 60 sessions. Whole blood samples were taken at the beginning of the study, at the midway point, and after the last session, followed by one last blood test a week or two after their last HBOT session. When the samples were viable, scientists tested the lengths of the subjects’ blood cell telomeres (the now-famous “shoestring nibs” on the ends of our chromosomes that keep them from unravelling after multiple cell divisions), as well as looking for what they characterize as “senescent” T-cells.
This is where the attrition problems started to get worse. After five of the initial recruits failed to complete their baseline survey, only 30 people remained to contribute samples — and of those, the researchers had to throw out four patients’ telomere analyses and ten patients’ “senescent” T-cell analyses, either because the samples contained too few cells to analyze, or because of lab technician errors. As such, the stated findings are based on very few data points indeed. And aside from reducing the statistical power to attribute (or not) any changes observed to random chance, counting only those data points may also have actively skewed the results. People who don’t complete a study or whose biological samples are not measurable could reflect real differences in the people who finish a trial versus those that began it: for instance, participants providing viable samples may have had unusually robust blood cells for their age, while those who completed baseline surveys may have been more conscientious (thus more inclined toward healthy lifestyle practices) than average.
Because of this potential source of bias, the accepted way to analyze a human clinical study is a so-called intention-to-treat analysis (ITT), in which you use all of the data from all of the subjects whether they actually finished the study (or gave viable samples) or not. Instead, in this case, they just ignored all the people who didn’t complete their initial assessment (that actually is sometimes acceptable, even in ITT), and also performed their analyses only on the people whose samples were all viable. To put it in fewer words, the researchers only compared people who started to the same people if they finished. Yet those who finished are skewed away from the whole group of those who started the study in the first place. Moreover, with no control group of any kind, we don’t know what would have happened anyway to another group of similarly-situated people who spent time under a kind of “placebo HBOT,” such as subjecting them to only very mild increases in atmospheric pressure, breathing something closer to atmospheric air (as).
So, all right: the meal comes with a cartoonishly-large pillar of salt on the side. But with all those caveats, what did they say they found?
After completing the protocol plus a two-week recovery period, telomere lengths in viable samples across several populations of immune cells “increased significantly by over 20% following HBOT,” and “There was a significant decrease in the number of senescent T-helpers by -37.30%±33.04 post-HBOT (P<0.0001)” while “T-cytotoxic senescent cell percentages decreased significantly by -10.96%±12.59 (p=0.0004)”. Is that actually what they found? And suppose that they had shown that (and shown it convincingly): would that be enough to justify a claim of having “stopped” or even “reversed aging”?
Tee-Tottering Telomeres
Let’s first look at the reported change in telomere length. It’s true that, when you look at a large population of people, longer blood cell telomeres and slower blood cell telomere shortening tend to correlate with poorer health outcomes. But that doesn’t make blood cell telomere length even a good proxy for current biological age at the individual level — let alone a causal driver of aging..
First, although blood cell telomere lengths do overall shrink over the course of multi-year periods when you look at aggregated data from an entire population of people, individuals’ telomere lengths swing wildly up and down over the course of mere months — that is, over lengths of time that include the entire duration of the HBOT study. So testing individual subjects’ telomere lengths as a measure of their individual biological age, and then testing again just a couple of months later, guarantees results that are corrupted by lot of sheer noise — and remember, just 24 subjects even had results the lab could read in the first place.
In fact, up to a third of individuals tracked show stable or even increased telomere lengths in their blood cells when re-tested as much as a decade later, due to a mixture of what are presumed to be real changes and lab artifacts. Obviously, degenerative aging is not being arrested or even massively reversed in one person in three across the population every ten years: imagine what an exciting breakthrough it would be to make that happen! So we know from that alone that we can’t use individual people’s blood cell telomere lengths as reliable indicators of age-related change, even over the course of a ten-year period. We therefore can’t possibly take a similar claim for HBOT seriously if it’s based on the same measure taken from 24 people over the course of mere months.
Moreover, although blood cell telomere length does correlate with telomere length in some tissues, the correlation is pretty weak, ranging from explaining 2% of the variation in the testes to a maximum of 14% in one peripheral nerve — and it has no correlation at all to the telomere lengths of about one-third of our tissues! Whatever blood cell telomere length tells us about health, in other words, it’s a pretty lousy proxy for whatever role telomere length plays in aging across the body as a whole.
Worse: calendar age itself — the most important determinant of biological age, which is what advocates want to use blood cell telomere length to measure — explained just 3.3% of the person-to-person variability in telomere length when all tissues were taken into account, and such important contributors to accelerating aging such as body mass index (BMI, as a proxy for obesity) and smoking status explained less than 1% of variation. And African Americans’ telomere lengths are longer than those of Americans of European descent in in nearly all tissues — a result consistent with multiple previous studies looking at blood cell telomere lengths. Yet we know that African Americans suffer with a higher burden of age-related disease and shorter life expectancies than white or Asian Americans. If we’re looking for a measure of biological aging, it makes little sense to use a metric that bears so little relationship to key predictors of future ill-health and death.
The weakness of telomere length as a biomarker of aging becomes clearer when you compare it to more robust markers, such as epigenetic aging clocks or algorithmic scores calculated mostly from common blood blood-test markers. In a comparative study, eleven candidate biomarkers of aging were compared to see how closely they would reflect the impact of aging on a group of older people in the domains of physical functioning (measured on tests of things like balance, grip strength, and motor coordination), rate of cognitive decline, and subjective signs of aging such as having an “old” face. One of the composite scores and one of the early iterations of an epigenetic aging clock consistently correlated with these age-related outcomes (albeit quite modestly in both cases), but telomere length failed to correlate with any of them. Similarly, telomere length was not associated with current health status in a cohort of 50-something New Zealanders, as measured on the SF-36 evaluation of health status.
And those first-generation epigenetic aging clocks are far inferior as predictors of future age-related morbidity and mortality than more recent iterations such as GrimAge and DNAm PhenoAge, as well as non-epigenetic biomarker composites such as the PhenoAge blood test composite score against which DNAmPhenoAge itself was trained up by machine learning. Indeed, a comparison study of aging Swedes found that eight out of nine different biological age scores predicted risk of death over the next twenty years beyond the predictive power of calendar age itself. Telomere length was the odd man out, being the only one that failed to add value to knowing a subject’s calendar age.
So that’s telomere length. What about the reduction in “senescent” T-cells?
There’s “Senescence” … and then there’s Senescence
Anyone who’s been following research on senescent cells and their roles in diseases of aging and age-related ill-health — and on the sweeping rejuvenation effects of triggering those cells to self-destruct with “senolytic” drugs — will be excited by the authors’ conclusion that their “study indicates that HBOT may induce significant senolytic effects” and “suggests a non-pharmacological method, clinically available with well-established safety profile, for senescent cells populations decrease.” But does the study actually support that claim?
The first thing to understand is that despite the fact that it’s standard terminology in the immunology world, “senescent” T-cells aren’t actually “senescent cells” in the sense usually used in the geroscience world. (And no, they aren’t anergic T-cells either, though the two do show some overlap). True senescent cells are in a state of total growth arrest, unable to make new copies of themselves due to an interlocking set of pathways of regulation. By contrast, “senescent” T-cells’ ability to proliferate is reduced, but still fundamentally intact. Additionally, the reasons for “senescent” T-cells’ short telomeres and reduced replicative ability are quite different from those of true senescent cells. T-cells are normally very “trigger-happy” with their telomere-lengthening telomerase enzyme (and possibly use another telomere-lengthening mechanism as well), which allows them to quickly replicate when they encounter a known target and flood the zone with new T-cells. “Senescent” T-cells’ telomere-lengthening activity is sluggish compared to normal T-cells because an energy-sensing pathway tunes down its access to the enzyme. By contrast, true senescent cells actively enforce a state of total growth arrest, using an interlocking set of cellular pathways that are just not active in normal cells, in order to prevent the replication of damaged, often cancer-prone cells.
In short, jumping from post-HBOT reductions in the number of these “senescent” T-cells to potential effects on classical senescent cells is really just a misunderstanding of what kinds of cells are involved in each case.
OK, you say, but still, these so-called senescent T-cells are bad, aren’t they? So getting rid of them must be good — right? Well, maybe (though no one has demonstrated that yet). But did HBOT actually get rid of them in the first place?
The first problem here is that we don’t even know for sure that the researchers were measuring “senescent” T-cells to begin with! T-cells are judged “senescent” based on blunted replicative ability, lack of the marker protein CD28 on their surface, and the abnormal presence of the marker CD57 — but the investigators here were only able to measure CD28, and therefore were judging T-cells “senescent” without actually using the full criteria to test for them. So whether there was even a change in “senescent” T-cells in the first place is quite uncertain. (The same goes, in fact, for natural killer (NK) cells, which are another kind of immune cells. Mature NK cells test negative for the marker CD3 but positive for the marker CD56 — but these investigators tested for cells positive for both markers! Maybe this was just a typo that got repeated throughout the manuscript, but as it stands, it can only further fuel uncertainty around the results as a whole).
Add to that the fact that a mere reduction in the numbers of measured cells doesn’t prove that HBOT destroyed them. Maybe HBOT somehow triggered this group of cells to adopt a different functional status. Or maybe the cells (or a subset of them) retreated back to their reservoirs in the lymph nodes, since the researchers were only sampling them in peripheral blood. Who knows?
What we certainly don’t know is that HBOT treatment destroyed these cells in substantial numbers — and again, even if it did, it would be unclear what the researchers had done, since they didn’t actually fully characterize these as “senescent” T-cells in the first place, and they got the result from just twenty samples (after throwing out ten unusable results) — all in a study with no control group.
“Aging Reversed”?
So in short, the actual details of the study show that even the narrow claims of the study abstract aren’t fully justified. It’s not clear that blood-cell telomeres were lengthened any more than they would have been without HBOT; it’s not clear that “senescent” T-cells were reduced in numbers, let alone actually destroyed; and if “senescent” T-cells had been destroyed, it would not demonstrate a senolytic effect of HBOT, because “those aren’t the ‘senescent’ cells you’re looking for.”
And even if the study had robustly demonstrated that every one of the points above really did occur, it would not constitute “reversing aging” — or even justify the more restrained claims that “blood cells actually grow younger as the treatments progress” or “that the aging process can in fact be reversed at the basic cellular-molecular level.”
Aging Reversed!
The mission of SENS Research Foundation is to accelerate the development of rejuvenation biotechnologies: new therapies that remove, repair, replace, and render harmless the real cellular and molecular damage of aging. Scientific studies have rigorously demonstrated that these proposed therapies can remove or repair members of the various categories of cellular and molecular aging damage in animal models — and in an increasing number of cases, in human clinical trials.
When we say that we aim to “reverse aging,” we actually mean reverse aging: not just that damaged cells and molecules will be removed and replaced, but that health and function will be restored — something the HBOT study did not even attempt to demonstrate. Bona fide rejuvenation can be seen (for example) in aging animals treated with senolytic drugs, and in humans with Parkinson’s disease given even the crude early forms of neuronal replacement therapy (and the first glimpses of its next generation).
Going forward, proof-of-concept studies at our Research Center and in expert labs funded by the Foundation will demonstrate more and more rejuvenation biotechnologies. The emerging rejuvenation biotechnology industry will continue to flourish, advance promising breakthroughs into human trials, and eventually license treatments for clinical use (first in the most at-risk, and increasingly in the otherwise-healthy aging). In a future we can see through a glass darkly, our vision will be revealed: a new humanity, open to an indefinite future free of the specter of decline. That day, we will see aging reversed.
