Weekly Issue — 2025-11-02

Quick gloss before we go further. An epigenetic clock isn't reading your DNA's letters; it's reading the chemical sticky notes — called methylation marks — that sit on top of your DNA and tell your cells which genes to use. As you age, the pattern of those sticky notes shifts in fairly predictable ways. Researchers built algorithms that look at the pattern and spit out a number: your estimated biological age, which can run younger or older than the candles on your cake.

Cool concept. But for years there's been a nagging worry: do these clocks actually mean the same thing in, say, Dublin as they do in Detroit? If the answer is no, paying for a test is basically buying a vibe. If the answer is yes, we might finally have a real-world dashboard for how we're aging.

A team led by Eric Klopack and colleagues pulled survey and blood data from three large aging studies — one in the United States, one in the Republic of Ireland, and one in Northern Ireland — and asked a simple question. When people smoke, drink heavily, or carry a higher BMI, do epigenetic clocks flag faster aging in the same way across all three places? The answer, published in Social Science & Medicine, was a striking yes — “remarkable consistency” across the studies.

Think of it like checking whether a thermometer reads the same in three different kitchens. If it does, you start to trust the thermometer. That's roughly the bar this paper clears for three behaviors we already know matter for long-term health.

Now, the careful part. This is one study. It's an observational analysis, which means it watches what people do and what their blood looks like — it doesn't randomly assign anyone to start smoking or stop drinking. That's why the editorial team here is calling the evidence moderate, not slam-dunk. Cross-country agreement is a real signal that these clocks are picking up something genuine about behavior and aging. It is not, on its own, proof that nudging your clock number will add years to your life.

The authors are upfront about the boundaries, too. All three samples are Western and relatively well-resourced. They explicitly call for more research in non-Western and low- and middle-income countries, and on health risks beyond the big three behaviors they tested. Translation: the thermometer looks reliable in three kitchens. We don't yet know how it reads in thirty.

Direct-to-consumer epigenetic-age tests are everywhere now, from wellness startups to longevity clinics. The honest framing, post-this-study, is that the underlying science is getting sturdier. The clocks aren't astrology. They're picking up real biological signal that correlates with behaviors we already know shape long-term health.

But here's the part the marketing tends to skip. A clock is a mirror, not a treatment. The same paper that strengthens the case for these tests also implies the obvious: if your clock moves, it's because something about you moved first — your smoking, your drinking, your weight. Buying a test and changing nothing is like weighing yourself more often and expecting the scale to do the work.

And there are things a single number genuinely can't tell you. It won't diagnose disease. It won't replace standard bloodwork. It won't account for sleep, stress, medications, or the dozens of other inputs researchers are still mapping. If you're considering a test — or already have results in hand — the responsible move is to bring them to a clinician who knows your full picture, not to DIY a protocol off a podcast.

Zoom out and the story here isn't really about gadgets. It's about a slow, careful tightening of the link between everyday behavior and measurable biology. Five years ago, “biological age” was a buzzword. Today, thanks to work like this cross-country analysis, it's looking more like a research tool with legs — one that, with more diverse data, could eventually help individuals and public-health teams track aging in a meaningful way.

For now, the most exciting thing about the clocks isn't what they promise. It's what they keep confirming. The things grandparents have been saying for a century — don't smoke, go easy on the booze, keep moving, eat like you love yourself — turn out to leave fingerprints on your DNA's sticky notes, in three countries and counting. The receipts are getting harder to argue with. What you do about them is still, refreshingly, up to you.

Dietary restriction — eating meaningfully less without malnutrition — is the most reproducible lifespan intervention in biology. The trouble is human. Sustained caloric restriction is brutally hard to maintain, and for women in midlife and beyond it carries its own risks around bone, muscle, and mood. So the search has long been for a molecule that flips the same biological switches without requiring anyone to push the plate away.

That's where rapamycin and metformin entered the chat. Both touch pathways — mTOR for rapamycin, AMPK and mitochondrial signaling for metformin — that overlap with the cellular response to eating less. But until now, the actual question, how does each compare to dietary restriction itself in vertebrates?, hadn't been answered head-to-head across the literature.

The new meta-analysis by Ivimey-Cook, Sultanova, and Maklakov in Aging Cell pooled 911 effect sizes from 167 papers covering eight vertebrate species. It is the most disciplined attempt yet to settle the comparison.

Three findings stand out. First, dietary restriction robustly extended lifespan across the vertebrate dataset, holding up across different ways of measuring lifespan and across different DR methodologies. That is reassuring rather than surprising — it confirms the benchmark.

Second, rapamycin produced a significant lifespan extension; metformin did not. In a field where both drugs are often invoked in the same breath, that is a meaningful divergence. It suggests that, at the vertebrate level, rapamycin is the molecule that more closely tracks the biology of eating less.

Third, sex didn't seem to consistently change the answer. Across treatments and across the way lifespan was reported, the authors did not find a reliable male–female split. For a literature that has historically over-relied on male mice, this is a small but useful note.

Here is where the language has to slow down. The authors themselves flag high heterogeneity and significant publication bias across treatments. Translated: studies varied widely in design and dose, and the published literature likely overrepresents positive results. The headline still survives those caveats — that is part of what makes meta-analysis useful — but it survives moderately, not triumphantly.

The results were also sensitive to how lifespan was reported (means versus medians, log-response framings). That doesn't erase the rapamycin signal, but it does mean a careful reader should resist any version of this story that reads as rapamycin extends life, full stop.

And then there is the species gap. Vertebrates include fish, rodents, and a handful of others — not humans. The conclusion the authors actually draw is bounded: rapamycin and dietary restriction confer comparable lifespan extension across a broad range of vertebrates. That is a statement about biology, not a prescription about you.

For readers whose physicians are already discussing geroprotectors, the practical reframe is this: the vertebrate evidence base now leans more strongly toward rapamycin as the molecule that imitates dietary restriction's lifespan effect. Metformin's case for general longevity use — as opposed to its well-established role in type 2 diabetes — looks weaker in this dataset than the cultural conversation has implied.

That is not the same as saying rapamycin is safe, appropriate, or beneficial for any individual woman in her fifties, sixties, or seventies. Rapamycin's human use to date is mostly in transplant medicine and oncology, at doses and schedules unlike anything contemplated for longevity. Its real-world side-effect profile in healthy adults, taken for decades, is genuinely not known. The honest position is curiosity with restraint.

For everyone else — which is most of us — the more durable takeaway sits upstream of the pills. Dietary restriction's lifespan signal is the most robust in the dataset, and the levers that approximate it without medication (protein-adequate caloric moderation, time-restricted eating windows under clinical guidance, resistance training to protect muscle and bone) remain the closest thing the field has to a free lunch. They are also, not coincidentally, the things a thoughtful clinician will discuss before any prescription is written.

What a meta-analysis like this really does is reset the prior. For years, rapamycin and metformin have been spoken of as siblings in the geroprotector conversation. After this paper, they look less like siblings and more like cousins with very different family resemblances to dietary restriction. That matters for which human trials deserve funding, which questions deserve our attention, and which claims — from clinics, podcasters, and supplement sellers — deserve a harder look.

The field is not done. Human outcomes data is what will ultimately matter, and we don't have it yet at the scale this question requires. But the vertebrate picture is now sharper than it was a year ago, and sharper is what serious readers came here for.

A meta-meta-analysis is exactly what it sounds like: a study of studies of studies. Researchers searched PubMed, SCOPUS and Web of Science through December 2024 for systematic reviews and meta-analyses on smoking and eye disease, graded each one with the 16-item AMSTAR 2 quality tool, and pooled the results. Sixteen reviews made the cut; twelve were strong enough to combine quantitatively.

Why bother layering analyses on top of analyses? Because individual meta-analyses on smoking and eyes have existed for years, and they don't always agree on the size of the risk. Stacking them is an attempt to find the signal across the noise — to ask, with more statistical horsepower than any single review, what the aggregate human evidence really says.

The short answer: it says smoking is bad for your eyes in ways that are larger, and more consistent across disease categories, than most of us casually assume.

Start with macular degeneration, the leading cause of irreversible central vision loss in older adults. In the pooled data, current smokers had an odds ratio of 11.93 (95% CI 4.40 to 32.33) and a risk ratio of 7.45 (95% CI 4.09 to 13.57) for AMD compared with people who never smoked. Translated: somewhere between a seven-fold and twelve-fold elevation, depending on how you slice the math. The confidence interval is wide — a reminder that the precise size of the effect is uncertain even if the direction is not.

Past smokers didn't get a clean pass. Their pooled odds ratio for AMD was 7.09 (95% CI 4.79 to 10.51) — still a roughly seven-fold increase relative to never-smokers. That is the part that tends to surprise readers: quitting moves the needle, but the retinal accounting doesn't reset overnight.

For primary open-angle glaucoma (POAG), the most common form of glaucoma, current smokers carried about three times the risk (OR 3.07, 95% CI 2.07 to 4.54), with past smokers at a similar 2.64. For cataracts — the clouding of the lens that eventually sends most of us to a surgeon — current smokers had roughly four times the risk (OR 4.15, 95% CI 3.35 to 5.15), and "ever" smokers about a six-fold increase.

Mechanistically, the eye is a small, metabolically demanding organ with delicate vasculature and tissues exquisitely sensitive to oxidative stress. Tobacco smoke delivers a cocktail of oxidants, constricts blood vessels, and is thought to compromise the antioxidant defenses that keep the retina and lens functioning over decades. None of that is new science. What's new — and useful — is the size of the aggregated human signal.

It is also worth saying what this study does not do. It does not run a randomized trial (you cannot ethically assign people to smoke). It does not finely separate dose, duration, vaping, or secondhand exposure. Observational evidence at this scale strongly suggests causation, but cannot prove it the way an RCT could. That is why the editorial evidence rating here is moderate, not definitive — the effect is consistent and large, but inherits the limits of the underlying studies.

Perimenopause is already a season of risk recalibration. Bone density, cardiovascular markers, sleep, mood — the dashboard lights up. Eye health rarely makes the top of that list, but it probably should, because the diseases in this analysis are largely silent until they aren't. AMD, glaucoma and cataracts develop over years; the choices that bend their trajectory are the ones you're making now.

If you currently smoke, the most honest read of this evidence is that cessation belongs alongside blood pressure and cholesterol on your healthspan shortlist — and that the eye-specific case for quitting is stronger than the public conversation suggests. If you used to smoke, the residual risk in the data is a reason to be proactive about dilated eye exams, not a reason to shrug. And if you've never smoked, this is one more argument for protecting that status, including from the underexamined risks of vaping and chronic secondhand exposure, which the review did not separately quantify.

None of this is medical advice, and none of it replaces a conversation with an ophthalmologist or your primary care clinician. But the synthesis does what good evidence reviews are supposed to do: it takes a vague cultural sense that smoking is "bad for your eyes" and gives it a number large enough to act on.

The evidence base on smoking and eye disease isn't new, but the resolution just got sharper. A 2025 meta-meta-analysis aggregates the best available reviews and lands on effect sizes — 7–12× for AMD, ~3× for glaucoma, ~4× for cataracts — that are too large and too consistent to dismiss as statistical noise, even with wide confidence intervals and the usual caveats of observational data. The direction of the action item is unambiguous: among the modifiable levers for protecting your vision into your sixties and seventies, quitting is still at the top of the list.