A failed Lp(a) trial and one that lowers LP(a) by 97%, a melanoma vaccine, the ketone your aging liver stops making, a gut bug that stretches healthspan, & more things worth your time this week.

Week of September 7th, 2026

The thread running through this week's research is subtraction: things your body quietly stops making, stops clearing, or starts making in the wrong place. We lead with a first-in-human trial in which a single injection knocked lipoprotein(a) down by up to 97% and held it there for nearly a year, then the phase 3 outcomes trial that spent seven years asking whether lowering that number actually prevents heart attacks and came back with an answer nobody wanted. From there: a Cell Metabolism paper showing that an aging liver stops producing the ketone your killer T cells need to stay in the fight, a two-cohort study asking whether the specific healthy diet you pick actually changes your epigenetic age, a gut bacterium that disappears with age and extends healthspan in mice when you put it back, the first structural map of how 40 different human tissues age at different speeds, a Nature paper naming an early Alzheimer's driver that produces the disease without any plaques at all, a nationwide Swedish study that quietly argues your colonoscopy guidelines are counting the wrong thing, and the first personalized cancer vaccine ever to clear a phase 3 trial. Nine findings, graded honestly. Here's what's worth your attention.

1. A Single Injection Cut Lipoprotein(a) by 97% for Nearly a Year

Lipoprotein(a) is the cardiovascular risk factor almost nobody gets tested for and nobody can diet away: it is set largely by the genes you were born with, it does not respond meaningfully to statins, and roughly one in five people carries an elevated level. In a first-in-human phase 1 trial published in The Lancet, Ashish Sarraju and Steven Nissen at the Cleveland Clinic, working with Chengdu Xinhua Hospital in China, randomized 71 otherwise healthy adults aged 18 to 55 with elevated lipoprotein(a) to a single subcutaneous dose of Kylo-11, a very long-acting small interfering RNA, or placebo across seven dose cohorts. Seventy received a dose and were followed a median of 337 days. At 48 weeks, median reductions in serum lipoprotein(a) ranged from 53% in the 9 mg cohort to 97% in the 600 mg cohort. In the cohort with baseline levels above 200 nmol/L, a single 225 mg dose produced a 96% median reduction, or 208 nmol/L in absolute terms. There were no serious adverse events, no injection-site reactions, and no adverse events the investigators attributed to the drug.

Grade it as a striking early signal that still has everything left to prove. The pharmacology here is remarkable: one shot, once a year, near-total suppression of a lifelong inherited risk factor. But this is a phase 1 safety study in 70 young, healthy volunteers with a median age of 27.5 years, and the primary endpoint was adverse events, not heart attacks. Lipoprotein(a) lowering remains a surrogate: no drug has yet demonstrated in a completed outcomes trial that driving this number down prevents cardiovascular events, and that question is exactly what the ongoing phase 3 programs in this class exist to answer. The practical takeaway for now is upstream of the drug entirely. If you have never had your lipoprotein(a) measured, it is a one-time test, it changes how aggressively you and your physician manage everything else, and this is a reasonable week to ask for it.

Sarraju A, Du X, Zhou L, et al. Safety and lipoprotein(a)-lowering effects of Kylo-11, a non-canonical, long-duration small interfering RNA targeting lipoprotein(a): a first-in-human, randomised, double-blind, placebo-controlled, phase 1 trial. The Lancet. 2026;408(10558):899-909. DOI: 10.1016/S0140-6736(26)01484-4. https://doi.org/10.1016/S0140-6736(26)01484-4

2. Your Aging Liver Stops Making the Ketone Your T Cells Need

One of the least discussed costs of aging is that the immune system loses its ability to kill things: aged CD8 T cells, the cells that clear virus-infected and cancerous cells, drift into a dysfunctional state called exhaustion, and nobody has been able to say precisely why. A team led by Jie Cheng at Huazhong University of Science and Technology and Peng Jiang at Tsinghua University, writing in Cell Metabolism, traced it to a metabolic supply problem. Aging represses BDH1, the liver enzyme that produces beta-hydroxybutyrate (3HB), the body's main circulating ketone. Less 3HB is made, less is taken up by T cells through the transporter SLC16A1, and the T cells fail. Deleting BDH1 in the livers of young mice reproduced the aged immune phenotype, compromising both antiviral and antitumor immunity, while supplementing 3HB reversed it. Mechanistically, 3HB chemically modifies a protein called PRKAR1B, which shuts off CREM, the transcription factor that switches on exhaustion genes. Most pointedly, CAR-T cells performed poorly in aged mice and were substantially restored by 3HB treatment.

Grade it as an unusually clean mouse mechanism with an obvious and unanswered human question attached. Every step of this is animal and cell work, so nothing here is a dosing recommendation. The reason it matters anyway is that the average cancer immunotherapy patient is in their sixties or seventies, precisely the population in which this circuit is failing, and the paper offers a cheap, well-characterized molecule as a potential fix. It also complicates the popular framing of ketones as something you achieve by cutting carbohydrates. The deficit here is in hepatic production, not dietary substrate, and whether swallowing exogenous ketone esters restores the same signal is genuinely unknown. Watch this one for a human trial in older immunotherapy patients rather than for a supplement label.

Xiao Y, Zhang Y, Zhao Z, et al. Aging microenvironment induces CD8 T cell exhaustion by suppressing hepatic beta-hydroxybutyrylate synthesis. Cell Metabolism. 2026. DOI: 10.1016/j.cmet.2026.08.009. https://doi.org/10.1016/j.cmet.2026.08.009

3. Ten Different Healthy Diets, One Shared Epigenetic Benefit

Nutrition arguments usually run on the assumption that one healthy pattern must be measurably better than the others, so it is worth knowing what happens when you test ten of them against a biological readout at once. Monique Breteler's group at the German Centre for Neurodegenerative Diseases in Bonn did exactly that in Nature Communications, scoring participants in the population-based Rhineland Study on ten established diet quality indices and validating the findings in the independent EPIC-Potsdam cohort. The first result is almost comic: there was minimal overlap in who landed in the top 25% of adherence across the different scores, meaning the diets are measuring genuinely different behavior. The second result is the useful one. Adherence to a healthy pattern was associated with reduced epigenetic aging across the board, though the magnitude varied by score, and while each dietary pattern carried its own distinct DNA methylation signature, those signatures largely converged on the same underlying biological pathways.

Grade it as a well-replicated association that quietly settles a common argument. This is observational and cross-sectional, epigenetic clocks are surrogate markers rather than outcomes, and people who eat well differ from people who do not in a hundred ways that statistical adjustment only partly handles. But the internal logic is what makes it persuasive: different diets, different methylation fingerprints, same destination. The honest translation is that the marginal return on picking the theoretically optimal diet is smaller than the return on actually adhering to any reasonable one. If you have been stalled between Mediterranean, DASH, and a plant-forward pattern of your own design, this is permission to stop optimizing and start complying.

Tavares JF, Liu D, Talevi V, et al. Associations between diet quality, epigenetic aging and epigenome in two population-based cohorts. Nature Communications. 2026;17(1). DOI: 10.1038/s41467-026-77064-4. https://doi.org/10.1038/s41467-026-77064-4

4. A Gut Bacterium That Disappears With Age, and Extends Healthspan When Restored

The microbiome has generated more longevity headlines than durable findings, largely because "the gut bacteria of old people look different" is a description rather than a lever. A large collaboration from the Chinese Academy of Sciences, led by Weiqi Zhang, Guang-Hui Liu, and Feng Zhang, went looking for the lever in Nature Aging. They first built a microbiome-based aging clock (MicroAge) to track biological aging trajectories, then used it to identify Bifidobacterium pseudocatenulatum as a species consistently depleted with age in both sexes across multiple Chinese cohorts. In naturally aged mice, oral administration of that single species restored intestinal homeostasis, reduced inflammatory signaling across multiple organs, improved cognitive and motor performance, and extended healthspan. They then identified the likely active ingredient: a bacterial metabolite called 5-aminovaleric acid betaine (5-AVAB), whose levels also decline in aging humans. Giving 5-AVAB alone partially reproduced the benefits, including the cognitive and motor gains and the suppression of multiorgan inflammation.

Grade it as a genuinely interesting target and a poor reason to buy a probiotic this week. The human component here is correlational, the intervention component is entirely in mice, and the endpoint is healthspan, meaning function and frailty, not lifespan. It is also worth knowing that B. pseudocatenulatum is not what is in your yogurt or in most commercial probiotic capsules, and that strain-level identity matters enormously in this field. What makes the work above average is the metabolite: identifying 5-AVAB converts a live-organism problem, which is hard to standardize and hard to regulate, into a small-molecule problem, which is neither. That is the version of this finding most likely to reach a clinic.

Lu X, Ping J, Han Z, et al. A geroprotective probiotic and its functional metabolite counteract inflammaging to extend healthspan. Nature Aging. 2026. DOI: 10.1038/s43587-026-01181-4. https://doi.org/10.1038/s43587-026-01181-4

5. The First Map of How 40 Human Tissues Age at Different Speeds

We talk about biological age as if the body had one clock, when the more accurate picture is forty clocks running at different rates. Sanju Sinha's group at Sanford Burnham Prebys, with collaborators at the National Cancer Institute and the NIH, built a framework called PathStAR that quantifies structural aging directly from routine histopathology slides, notably without being trained to predict chronological age. Published in Nature Aging, they applied it to 25,306 post-mortem biopsies spanning 40 tissues from 970 donors aged 21 to 70. Organs turned out to age along distinct, nonlinear trajectories: vascular tissue accelerates early, uterine and vaginal tissue accelerates late and around menopause, and digestive and male reproductive organs show two separate bursts. Across organs, the periods of acceleration shared a common signature of increased inflammation alongside reduced energy production, repair, and quality control. Deterioration was also coordinated within individuals, with digestive and male reproductive tissues linked through sex hormones.

Grade it as a map rather than a mile marker. This is cross-sectional post-mortem tissue, not a longitudinal study, the donors top out at 70, and PathStAR is a research tool, not a test you can order. Its value is in the shape of the curves. The finding that vascular structural aging accelerates earliest is the one with teeth, because it puts the fastest-moving damage in the decade when most people consider themselves too young to bother, and blood pressure, lipids, lipoprotein(a), and glucose are all modifiable in your thirties and forties in a way that vessel wall architecture is not in your seventies. The second useful idea is coordination: organs do not age independently, so a single system going off the rails is worth treating as a systemic signal rather than a local problem.

Yadav A, Alvarez K, Chechenina A, et al. Mapping structural aging across human tissues reveals tissue-specific trajectories and coordinated deterioration. Nature Aging. 2026. DOI: 10.1038/s43587-026-01200-4. https://doi.org/10.1038/s43587-026-01200-4

6. An Alzheimer's Driver That Produces the Disease Without Any Plaques

The amyloid hypothesis has absorbed most of the field's attention and most of its failed trials, so a paper that reproduces core Alzheimer's pathology with no plaques at all deserves a careful read. Won-Suk Chung's group at KAIST and the Institute for Basic Science in Korea, publishing in Nature, found that as disease progresses in Alzheimer's mouse models, astrocytes and microglia ramp up their elimination of excitatory synapses while pruning fewer inhibitory ones. Single-nucleus RNA sequencing then identified one of the earliest major changes: a population of "early-responsive excitatory neurons" defined by ectopic expression of the receptor ERBB4. Selectively deleting Erbb4 from excitatory neurons in these mice abolished the abnormal network activity, the synapse loss, the reactive gliosis, the amyloid plaque deposition, and the cognitive deficits. Running it in reverse, overexpressing Erbb4 in the excitatory neurons of healthy mice reproduced those same Alzheimer's-like features without producing amyloid plaques. The effects required mTOR signaling downstream, and a mediation analysis of human Alzheimer's transcriptomic data was consistent with the same cascade.

Grade it as a mechanistic result that reorders the suspects rather than a therapy. This is mouse genetics plus inference from human tissue transcriptomes, with no drug and no trial, and ERBB4 does necessary work elsewhere in the nervous system, so blanket inhibition is not obviously safe. What earns it a spot this week is the reverse experiment. Producing network dysfunction, synapse loss, gliosis, and memory failure in an animal with no plaques is a direct argument that plaque burden is one output of the pathology rather than its engine, which is consistent with why plaque-clearing drugs have delivered such modest clinical effects. There is nothing here to act on, and there is something here worth updating on.

Lee SY, Park E, Lee HE, et al. Aberrant excitatory neuronal ERBB4 promotes Alzheimer's disease pathology. Nature. 2026. DOI: 10.1038/s41586-026-10964-z. https://doi.org/10.1038/s41586-026-10964-z

7. Your Colon Cancer Guidelines May Be Counting the Wrong Relatives

Screening guidelines routinely single out family history of early-onset colorectal cancer as the trigger for closer surveillance, which is a reasonable-sounding rule that had not been rigorously tested against the alternative. A Karolinska Institutet team led by Asa Everhov and Ola Olen tested it in Gastroenterology using Swedish nationwide registers from 1996 to 2023, comparing 124,387 patients with inflammatory bowel disease against 1,213,641 individuals matched on age, sex, parish, and year. Over a median 11 years of follow-up there were 1,882 colorectal cancers in the IBD group, an incidence rate of 1.17 per 1,000 person-years, versus 14,177 cancers and a rate of 0.88 in comparators. The decisive numbers are in the family-history breakdown. Having two or more affected first-degree relatives was associated with 2.69 additional cancers per 1,000 person-years, while having a relative diagnosed before age 50 was associated with only 0.42 additional cases, a confidence interval that crossed zero. On the absolute scale, family history raised risk similarly in IBD patients and in the general population.

Grade it as high-quality observational data that argues, carefully, against a piece of current guidance. It is registry-based rather than randomized, it is a Swedish population that may not generalize cleanly, and showing that a group is higher risk is not the same as showing that surveilling them more often saves lives. But the authors' point stands on its own arithmetic: guidelines emphasize the age at which a relative was diagnosed, and this dataset says the count of affected relatives carries more absolute weight. If you have two or more first-degree relatives with colorectal cancer, that is worth raising explicitly with your gastroenterologist, whether or not any of them were diagnosed young.

Everhov AH, Kristjansson K, Ludvigsson JF, et al. Impact of Family History on Colorectal Cancer Risk in Inflammatory Bowel Disease and in Matched General Population Comparators. Gastroenterology. 2026. DOI: 10.1053/j.gastro.2026.08.029. https://doi.org/10.1053/j.gastro.2026.08.029

8. The First Large Lp(a) Outcomes Trial Missed Its Primary Endpoint

Nothing in this week's collection cuts against another item more directly than this one. Four days before we published, Novartis announced topline results from Lp(a)HORIZON, the first phase 3 cardiovascular outcomes trial ever built to answer whether lowering lipoprotein(a) actually prevents cardiovascular events. It did not. The trial randomized 8,323 patients with established cardiovascular disease and lipoprotein(a) at or above 70 mg/dL to monthly subcutaneous pelacarsen, an antisense oligonucleotide that shuts down apolipoprotein(a) production in the liver, or to placebo. The primary endpoint was four-point MACE: cardiovascular death, non-fatal myocardial infarction, non-fatal stroke, and urgent coronary revascularization requiring hospitalization, assessed in the overall population and in a prespecified subgroup at or above 90 mg/dL. Pelacarsen lowered lipoprotein(a) substantially, consistent with the reductions of up to 80% seen in phase 2. That lowering did not translate into fewer cardiovascular events. Safety was described as acceptable. That is the entirety of what has been released, and the full dataset goes to a medical congress at an unspecified future date.

Grade it as preliminary, unpublished, and considerably less conclusive than the headline. Everything above comes from a press release: there is no hazard ratio, no confidence interval, no event count, no separation curve, no subgroup result. Three structural features of the trial deserve attention before anyone concludes that lipoprotein(a) does not matter. First, this is a lifelong exposure being treated for a few years. Lipoprotein(a) is roughly 90% genetically determined and elevated from childhood, and the genetic evidence that it causes atherosclerosis comes from people who have carried that burden for five or six decades. Mendelian randomization work has consistently suggested that the absolute reduction required to move event rates inside a short trial window is very large, which is a different question from whether the molecule is causal. Second, established disease may simply be the wrong target. Every participant already had built plaque, and removing one contributor to a wall that has already gone up is a much harder ask than never building it. Third, and most limiting, these were optimally treated patients, all receiving guideline-directed care including lipid-lowering and antihypertensive therapy. The better everything else is managed, the smaller the residual risk left for lipoprotein(a) to explain, and the smaller the signal any trial can detect. What this result narrowly answers is whether this drug, in this population, at this stage of disease, over this duration, changes outcomes. Two differently designed siRNA outcomes trials are still running: OCEAN(a)-Outcomes requires lipoprotein(a) at or above 200 nmol/L, and ACCLAIM-Lp(a) enrolls people at risk of a first event rather than only those who already have disease, with a further Amgen trial moving earlier still. The advice in item 1 stands unchanged, for a slightly modified reason. Measure it once, and use the number to decide how hard to push on the risk factors you can actually move today.

Novartis. Novartis announces Lp(a)HORIZON Phase III topline results for pelacarsen in patients with elevated Lp(a) and established cardiovascular disease (CVD). Media release. September 4, 2026. https://www.novartis.com/news/media-releases/novartis-announces-lpahorizon-phase-iii-topline-results-pelacarsen-patients-elevated-lpa-and-established-cardiovascular-disease-cvd

Cho L, Nicholls SJ, Nordestgaard BG, et al. Design and Rationale of Lp(a)HORIZON Trial: Assessing the Effect of Lipoprotein(a) Lowering With Pelacarsen on Major Cardiovascular Events in Patients With CVD and Elevated Lp(a). American Heart Journal. 2025;287:1-9. DOI: 10.1016/j.ahj.2025.03.019. https://doi.org/10.1016/j.ahj.2025.03.019 9. A Personalized Cancer Vaccine Cleared Phase 3, With the Numbers Still Sealed

Therapeutic cancer vaccines have been promising for roughly thirty years and delivering for almost none of them, which is why the August announcement matters more than the usual press release. Merck and Moderna reported that INTerpath-001, a phase 3 trial in 1,137 patients with completely resected stage IIB through IV cutaneous melanoma and no prior systemic therapy, met its primary endpoint of recurrence-free survival and its key secondary endpoint of distant metastasis-free survival at a prespecified interim analysis. Patients were randomized to intismeran autogene, formerly mRNA-4157 or V940, plus pembrolizumab, or to pembrolizumab alone, which is the current standard of care. Both improvements were described as statistically significant and clinically meaningful, with no new safety signals, and the trial continues in order to collect overall survival. This is the first positive phase 3 readout for an individualized neoantigen therapy and for any mRNA-based cancer therapy. The approach is genuinely bespoke: the resected tumor is sequenced against the patient's own germline DNA, up to 34 mutations predicted to be visible to the immune system are selected, and all of them are encoded into a single mRNA construct manufactured for that one person. It builds on the phase 2b KEYNOTE-942 study, published in the Journal of Clinical Oncology in June, where among 157 patients with resected stage IIIB to IV disease randomized 2:1, at a median 60.3 months of follow-up the combination showed a recurrence-free survival hazard ratio of 0.510 (95% CI, 0.294 to 0.887) and a distant metastasis-free survival hazard ratio of 0.411 (95% CI, 0.200 to 0.843).

Grade it as the most consequential oncology result of the year and, at this moment, a press release with no numbers in it. There is no hazard ratio, no confidence interval, no absolute recurrence rate, no median follow-up, and no subgroup breakdown. Two statistical cautions are worth carrying into the eventual presentation. Interim analyses that cross an efficacy boundary systematically overstate the true effect size, so the published hazard ratio should be expected to look more modest than the topline language implies. And recurrence-free survival, while an accepted regulatory endpoint in adjuvant melanoma, is not survival: the five-year phase 2b data are the honest illustration, where the overall survival hazard ratio was 0.471 with a confidence interval of 0.165 to 1.345, an interval that crosses one, built on seven deaths in each arm. The widely quoted five-year survival figures of 92.2% versus 71.3% rest on those same fourteen events. It is also worth noting that the phase 3 expanded eligibility down to stage IIB and IIC, a lower-risk group where the absolute benefit of any adjuvant therapy is necessarily smaller than in stage III or IV. The practical constraint is manufacturing rather than biology: every dose requires tumor sequencing, neoantigen prediction, and an individual production run, which means weeks of turnaround and a cost and logistics structure that no oncology practice currently has in place. Nothing here is actionable this week, because the therapy is not approved. What earns it the last slot is that the platform is now being tested in non-small cell lung, renal cell, and bladder cancer, and this readout is the first evidence that the underlying idea works at all.

Merck & Co, Moderna Inc. Merck and Moderna Announce Phase 3 INTerpath-001 Trial of Intismeran Autogene Plus KEYTRUDA Met Endpoints of Recurrence-Free Survival (RFS) and Distant Metastasis-Free Survival (DMFS) in Patients With Completely Resected Stage IIB-IV Melanoma. Media release. August 19, 2026. https://www.merck.com/news/merck-and-moderna-announce-phase-3-interpath-001-trial-of-intismeran-autogene-plus-keytruda-met-endpoints-of-recurrence-free-survival-rfs-and-distant-metastasis-free-survival-dmfs-in-patient/

Khattak A, Carlino MS, Meniawy T, et al. Intismeran Autogene Plus Pembrolizumab Versus Pembrolizumab Alone in High-Risk Resected Melanoma: 5-Year Update of the Randomized Phase IIb KEYNOTE-942 Study. Journal of Clinical Oncology. 2026;44(26):2482-2489. DOI: 10.1200/JCO-26-00835. https://doi.org/10.1200/JCO-26-00835

Happy Labor Day,

Dr. Smith and Winston

Medical disclaimer

The content above is for informational and educational purposes only and does not constitute medical, nutritional, or professional advice. It is not intended to diagnose, treat, cure, or prevent any disease and is not a substitute for personalized guidance from a licensed physician or other qualified healthcare provider. Individual needs and risks vary. Do not begin, stop, or modify any exercise regimen, dietary plan, medication (including lipid-lowering or lipoprotein(a)-lowering agents, immunotherapies, or investigational cancer therapies), or supplement (including ketone esters or probiotics). Consult your doctor first — particularly if you have or suspect a cardiovascular, metabolic, kidney, gastrointestinal, or neurological condition, take prescription medications, or are pregnant or breastfeeding. Studies summarized here vary in design and quality; preclinical and animal findings often do not translate to humans, and associations reported in observational research do not prove cause and effect. Studies summarized here vary in design and quality; preclinical and animal findings often do not translate to humans, associations reported in observational research do not prove cause and effect, and topline results announced by manufacturers before peer-reviewed publication may change once full data are presented.